Succession, Replacement, and Modification of Chicken Litter Microbiota

ABSTRACT

Chickens are in constant interaction with their environment, e.g., bedding and litter, and their microbiota. However, how litter microbiota develops over time and whether bedding and litter microbiota may affect the cecal microbiota is not clear. We addressed these questions using sequencing of V3/V4 variable region of 16S rRNA genes of cecal, bedding, and litter samples from broiler breeder chicken flocks for 4 months of production. Cecal, bedding, and litter samples were populated by microbiota of distinct composition. The microbiota in the bedding material did not expand in the litter. Similarly, major species from litter microbiota did not expand in the cecum. Only cecal microbiota was found in the litter forming approximately 20% of total litter microbiota. A time-dependent development of litter microbiota was observed. Escherichia coli, Staphylococcus saprophyticus, and Weissella jogaejeotgali were characteristic of fresh litter during the first month of production. Corynebacterium casei, Lactobacillus gasseri, and Lactobacillus salivarius dominated in a 2-month-old litter, Brevibacterium, Brachybacterium, and Sphingobacterium were characteristic for 3-month-old litter, and Salinococcus, Dietzia, Yaniella, and Staphylococcus lentus were common in a 4-month-old litter. Although the development was likely determined by physicochemical conditions in the litter, it might be interesting to test some of these species for active modification of litter to improve the chicken environment and welfare.

IMPORTANCE Despite intimate contact, the composition of bedding, litter, and cecal microbiota differs considerably. Species characteristic for litter microbiota at different time points of chicken production were identified thus opening the possibility for active manipulation of litter microbiota.

INTRODUCTION

Humans and animals are in continuous interaction with their environment. Eukaryotic hosts are exposed to microbiota in the feed, and another subset of microbiota may interact with humans or animals via drinking water, air dust, or flooring. Conversely, eukaryotic hosts release a large amount of their microbiota back into the environment, particularly through their fecal microbiota. However, studies analyzing interactions between environmental microbiota and host gut microbiota are not common despite their importance for human and animal health and performance. What happens to fecal microbiota when stool or feces are released into the environment? Do these dominate in the litter or is litter microbiota of a specific composition?

Chickens in commercial production are raised on flooring covered with bedding material such as straw, sawdust, or shredded paper (1). Bedding material can, therefore, act as the first source of microbiota for chickens after their placement on the farm. Later on, fecal droppings are mixed with the original bedding forming a new substrate, which acts as an ecological niche, with specific characteristics, such as increased amounts of organic molecules excreted by chickens, remains of the feed spread over the floor, and continuous inoculation by bacteria colonizing chickens (2). This gradually increases osmolarity, moisture, and pH (3, 4) or the appearance of microecological niches with reduced oxygen access. Because this material is usually not removed during chicken production, its microbial composition may change over time as additional fecal material is deposited. In some production practices, litter is used repeatedly as bedding material (5, 6). A chicken litter with its microbiota is, thus, an important material for which chickens are in intimate and permanent interaction.

Bedding material and litter can be a reservoir of commensal gut microbiota, and pathogens or may contain their specific microbiota. Bedding material and litter microbiota can be also reservoirs of horizontally transmissible antibiotic-resistance genes (7). Because fecal material is permanently deposited, one can expect that fecal microbiota will be present also in the litter. On the other hand, the litter environment is an aerobic one preventing the replication of the majority of gut microbiota. However, there are facultative anaerobes, aerotolerant or spore-forming bacteria among gut microbiota that might be able to cope with the conditions in the litter. There have been several reports analyzing litter microbiota. These studies indicate that if gut and litter microbiota is compared at the phylum level, both contain Gram-positive Firmicutes and Actinobacteria (8–10). However, at lower taxonomic levels, litter microbiota differs from the gut microbiota of poultry. Litter microbiota members from phylum Firmicutes are represented mainly by the family Staphylococcaceae and those from phylum Actinobacteria by families Brevibacteriaceae, Dermabacteraceae, Dietziaceae, Micrococcaceae, and Corynebacteriaceae, which are all rare in chicken gut microbiota (11). However, these studies were performed in broilers, i.e., for 40 days only (5, 8–10). Moreover, the microbiota of the original bedding was never compared with gut or litter microbiota.

In this study, we were interested in the extent that bedding microbiota entered the cecal or litter microbiota and in the extent that litter and cecal microbiota interacted. In addition, we tested the succession and replacement of bacterial species in the litter microbiota for 4 months, i.e., longer than in earlier studies performed in broilers mostly up to 1 month of age (5, 8–10). We also tested to what extent litter microbiota acted as a reservoir of mobile antibiotic resistance genes common in gut microbiota. Understanding the composition and interactions of bedding, chicken, and litter microbiota may help in the design of novel biological products enabling control of environmental microbiota at farms and improving animal health and welfare.

RESULTS

Sequence data used in downstream analyses.

In total, 5,249,009 reads were obtained for 162 analyzed samples with an average coverage of 32,401 reads per sample. Minimal and maximal sample read coverages were 1,223 and 120,078 reads, respectively. There were 6 samples with coverage lower than 10,000 reads, 41 samples with 10,001 to 20,000 read coverage, 71 samples with 20,001 to 40,000 read coverage, 29 samples with 40,001 to 60,000 read coverage, and 15 samples with coverage higher than 60,000 reads. This meant that we should have detected bacteria forming 0.01% of the total population but missed those that had lower representation. Whenever there was reporting on the absence of a particular bacterium, this was understood as absent or below the detection limit of 0.01% of the total population.

Major characteristics and differences of bedding, cecal, and litter microbiota.

Original bedding gradually developed into a litter cake, the top of which was covered with solid material. However, the solid cover material at the top was frequently broken in irregular places, thus allowing access to air and oxygen to the material beneath the litter cake.

Permutational analysis of variance (PERMANOVA) showed that there were significant differences in microbiota composition among all analyzed samples, except for microbiota in 3- and 4-month-old litters (Table 1 and Fig. 1A). Clean bedding microbiota consisted mainly of classes Bacteroidia, Gammaproteobacteria, and Alphaproteobacteria with Alphaproteobacteria being the most characteristic for straw bedding microbiota (Fig. 1B). At the order level, bedding microbiota was dominated by Rhizobiales and Sphingomonadales (both Alphaproteobacteria), Pseudomonadales and Burkholderiales from Gammaproteobacteria and Flavobacteriales, Sphingobacteriales and Cytophagales from Bacteroidia (Fig. 1C).

TABLE 1.

Similarities and differences in cecal and litter microbiota composition as determined by PERMANOVA^a

Material	Bedding	Cae_M1b	Cae_M4	Litt_M1	Litt_M2	Litt_M3	Litt_M4
Bedding		P < 0.01	P < 0.01	P < 0.01	P < 0.01	P < 0.01	P < 0.01
Cae_M1	P < 0.01		P < 0.01	P < 0.01	P < 0.01	P < 0.01	P < 0.01
Cae_M4	P < 0.01	P < 0.01		P < 0.01	P < 0.01	P < 0.05	P < 0.05
Litt_M1	P < 0.01	P < 0.01	P < 0.01		P < 0.01	P < 0.01	P < 0.01
Litt_M2	P < 0.01	P < 0.01	P < 0.01	P < 0.01		P < 0.01	P < 0.05
Litt_M3	P < 0.01	P < 0.01	P < 0.05	P < 0.01	P < 0.01		NS
Litt_M4	P < 0.01	P < 0.01	P < 0.05	P < 0.01	P < 0.05	NS

^aNS, not significant.

^bCae, microbiota composition in the cecum; Litt, microbiota composition in the litter; M1 to M4, age of the chicken flock and/or litter development in months.

FIG 1.

Bedding, litter, and cecal microbiota. Bedding, cecal and litter microbiota significantly differed from each other. Litter microbiota stabilized between months 3 and 4 resulting in the absence of difference by PERMANOVA (P > 0.05). (A) Multidimesional scaling (MDS) of bedding, litter, and cecal samples. (B and C) Microbial composition of individual types of samples at the class and order level, respectively. Only major taxa were identified. For full microbiota composition, see Table S1 in Supplemental File 1.

Cecal microbiota was characterized by Bacterodia and Clostridia (Fig. 1B), which comprised orders Bacteroidales, Oscillospirales, Lachnospirales, and Clostridia vadinBB60 group (Fig. 1C).

Litter microbiota was characterized by the presence of Gammaproteobacteria (order Enterobacterales), Actinobacteria (orders Corynebacteriales and Micrococcales), and Bacilli with orders Staphylococalles, Lactobacillales, and Bacillales. Time-dependent development of microbiota composition was recorded in the litter with Enterobacterales and Lactobacillales being present in the litter microbiota during the first month and disappearing from litter microbiota during the second and third months, respectively (Fig. 1B and C).

Microbial taxa specific for clean straw bedding, cecum, and litter at different stages of development.

Detailed identification of taxa characteristics for each type of sample was performed by linear discriminant analysis effect size (LEfSe) limited to amplicon sequence types (ASVs), which formed at least 0.25% of the total microbiota in at least one type of sample. A total of 202 ASVs passed this criterion. Because many ASVs were not identified by Qiime down to species, the 16S rRNA sequence of 202 ASVs was manually annotated using the rRNA/internal transcribed spacer (ITS) database in GenBank. After the annotation, LEfSe without classification into higher taxa was performed (Fig. 2A). The most specific bedding microbiota members included Sphingomonas olei, Sphingomonas echinoides, Rhodococcus fascians, and Rhizobacterium soli. Of potential pathogens, Bordetella bronchiseptica formed 0.6% of bedding microbiota. The cecal microbiota of chickens younger than 1 month was dominated by Bacteroides caecicola, Bacteroides fragilis, and Ruminococcus torque (Fig. 2A). Four-month-old chickens were colonized by Bacteroides gallinaceum, Acetobacteroides hydrogenigenes, and Mucispirillum schaedleri. Litter microbiota during the first month of chicken production contained Escherichia coli, Staphylococcus saprophyticus, and lactic acid bacteria (Weissella jogaejeotgali, Lactobacillus gallinarum, and Enterococcus faecium). These were replaced by Corynebacterium casei, different lactobacilli species, and Staphylococcus piscifermentans in the second month. The abundance of lactobacilli in the litter microbiota decreased during the third month of chicken production and these were replaced by Brachybacterium faecium, Brevibacterium iodinum, or Sphingobacterium lactis. During the fourth month of chicken production, Salinicoccus kunmingensis, Staphylococcus lentus, Dietzia aerolata, Oceanobacillus chironomi, Brevibacterium metallicus, or Yaniella halotolerans increased to their maximal abundance (Fig. 2A and B).

FIG 2.

Species-specific for microbiota of tested samples. LEfSe identification of species significantly enriched in each type of sample (A). Time-dependent abundance of the top 3 ASVs from each type of sample (B) also documented that major bedding and litter microbiota members did not replicate in the chicken intestinal tract. Although, chicken cecal microbiota could be detected in the litter.

Cecal microbiota and their persistence in the litter.

Data in Fig. 2B showed that cecal microbiota members survived to a different extent in the litter environment. Next, we analyzed the presence of gut microbiota DNA in the litter in greater detail. This analysis was performed with ASVs which formed at least 0.1% of the total microbiota in the chicken cecum up to 1 month of age. The criterion was passed by 140 ASVs which formed 80.38% of total cecal microbiota and 38.63% of 1-month-litter microbiota (Fig. 3A). Of these, E. coli, L. gallinarum, Lactobacillus gasseri, and Limosilactobacillus reuteri formed 11.44, 5.17, 3.2 and 1.78% of litter microbiota during the first month of rearing chickens. When the facultative anaerobes and aerotolerant species were excluded, strict gut anaerobes formed 17.07% of the month 1 litter microbiota. The same analysis performed for the cecal microbiota of 4-month-old chickens showed that the top 152 ASVs from cecal microbiota formed 77.63% of total cecal microbiota and 21.57% of litter microbiota (Fig. 3C).

FIG 3.

Environmental survival of the most abundant gut microbiota members. Except for facultative anaerobes (E. coli in blue; L. gallinarum, L. gasseri, and L. reuteri in different shades of green in [A and B]), gut anaerobes formed approximately 20% of litter microbiota ([A and C] for month 1 and month 4, respectively). After normalization of the top 30 cecal ASVs to 100% (B and D), ratios for each ASV of its cecal and litter abundance were calculated and plotted as log₂ ratios in (E and F) for 1- and 4-month-old chickens and their litter environment, respectively. The most common ASVs highlighted in (B and D) were (1) Bacteroides caecigallinarum, (2) Bacteroides fragilis, (3) Bacteroides caecicola, (4) Mediterranea massiliensis, (5) Parabacteroides distasonis, (6) [Ruminococcus] torques, (7) Escherichia coli, (8) Kineothrix alysoides, (9) L. gallinarum, (10) Barnesiella viscericola, (11) Porphyromonas pogonae, (12) Bacteroides gallinaceum, (13) Acetobacteroides hydrogenigenes, (14) Ruficoccus amylovorans, (15) Bacteroides caecigallinarum, (16) Barnesiella viscericola, (17) Bacteroides caecigallinarum, (18) [Ruminococcus] torques, (19) Mucispirillum schaedleri, (20) Bacteroides caecigallinarum. For full taxonomy, see Table S1 in Supplemental File 1.

The data in Fig. 3A and C were affected by the rest of the litter microbiota. In the next step, we compared the abundance of the top 30 ASVs from cecal microbiota with their normalized abundances in the litter (Fig. 3B and D), which were then used for the calculation of ratios. Besides E. coli and L. gallinarum, of the other cecal microbiota members, Faecalibacterium prausnitzii, Bacteroides caecicola, and Bacteroides thetaiotaomicron persisted the best in the litter during the first week of chicken production. On the other hand, Barnesiella viscericola or Odoribacter splanchnicus did not persist in the environment at all (Fig. 3E). The accumulation of fecal material resulted in improved environmental persistence of gut microbiota DNA in month 4 compared with month 1 because except for Mucispirillum schaedleri, the ratio of the remaining bacteria in the cecal and litter microbiota was close to a value of 1 (Fig. 3F).

Litter microbiota as a reservoir of antibiotic resistance genes.

Here, we investigated whether litter microbiota could act as a reservoir of antibiotic-resistance genes common to gut microbiota. Each of the target genes represented a marker of a particular taxon. bla_TEM lactamase was selected as common in E. coli and Enterobacteriaceae. tetO is a broadly distributed gene detected in Gram-negative Campylobacter or Fusobacterium (12) as well as in Gram-positive isolates from the family Lachnospiraceae (13). tetQ is characteristic of Bacteroidetes (14), and tetW is common to Lachnospiraceae and Ruminococcaceae but can spread to Lactobacilli and Actinobacteria as well (13). tet(44) is associated with Erysipelotrichaceae and tetA(P) we earlier detected in low GC content Clostridia such as Clostridium perfringens (13). tetQ, tetW, and tet(44) were more frequent in cecal microbiota than in the litter microbiota indicating their linkage with gut microbiota and release into the environment with fecal material. The tetO gene was as frequent in the litter microbiota as in the cecal microbiota and bla_TEM and tetA(P) were more frequent in litter microbiota than in the cecal microbiota (Fig. 4). None of these genes was detected in microbiota from bedding, indicating an no or low contribution of this type of material to the spread of antibiotic resistance among chicken gut microbiota (Fig. 4). Correlation analysis linked bla_TEM with E. coli. tetW was most positively correlated with the presence of Flintibacter butyricus, tet(44) positively correlated with Bacillus coahuilensis, tetO positively correlated with Kineothrix alysoides and tetA(P) positively correlated with Romboutsia timonensis (Fig. 5).

FIG 4.

Distribution of selected antibiotic resistance genes in bedding, litter, and cecal microbiota. bla_TEM and tetA(P) were more common in the litter microbiota while tetQ, tetW, and tet(44) genes were more frequent in cecal microbiota than in the litter or bedding microbiota. tetO was similarly distributed in the cecal and litter microbiota.

FIG 5.

Correlation of antibiotic resistance genes in tested samples and microbiota composition. Spearman’s correlation (expressed as a heat map) was used to predict the species most likely acting as a reservoir of specific antibiotic resistance genes.

DISCUSSION

In this study, we analyzed the development of microbiota in the chicken litter for 4 months. The limit of this study is that only sequencing data were used. We, therefore, could not differentiate between DNA originating from live or inactivated bacteria. In addition, PCR amplification of the eubacterial 16S rRNA gene was used before sequencing and, therefore, we completely lost information on yeast, fungi, molds, or protozoa in the litter despite their common presence in this material (4, 15).

Immediately after chicken placement on the farm, the litter environment of chickens in commercial production began to develop. Within a few days, the original microbiota of the bedding decreased below detection limits and was replaced by the microbiota of the forming litter. Bedding microbiota neither entered the chicken intestinal tract nor contributed to the formation of litter microbiota, although a low number of bedding samples were analyzed, leaving such a conclusion valid only for this study. Similarly, the major litter microbiota members did not replicate in the gastrointestinal tract, as noted earlier (5). Only cecal microbiota could be found in litter, which was not surprising because fecal droppings are continuously excreted by chickens into their environment. During the first month, E. coli, lactobacilli, and, unexpectedly, Faecalibacterium prausnitzii exhibited higher abundance in the litter than in the cecum. Using DNA sequencing only, we cannot decide on the viability of F. prausnitzii. However, the ability of F. prausnitzii DNA to persist in the environment was recorded earlier in experiments with reused litter (5, 8) and may be associated with its loss of spore formation (16) and evolution of new survival strategies in aerobic conditions by scavenging low levels of oxygen (17, 18). Except for E. coli, lactobacilli, and Faecalibacterium, DNA from the remaining cecal microbiota members persisted poorly in the litter during the first month of chicken production. On the other hand, conditions in the litter “cake” after 4 months were more favorable for gut microbiota survival because the ratio of the microbiota members in the cecum and litter was around a value of 1 except for Mucispirillum, which colonizes mucus and not intestinal volume and digesta (19). A higher abundance of gut microbiota DNA was likely caused by the solid nature of the litter cake in month 4, locally providing microniches with limited air exposure. Local microenvironments with limited oxygen access should not be confused with strictly anaerobic conditions, which would allow gut anaerobes to multiply. Strict gut anaerobes likely did not multiply in the litter because these formed only around 20% of the total litter microbiota despite their continuous deposition.

The chicken was the most important factor influencing the formation of litter microbiota. Similar to the results of De Cesare et al. (8), E. coli appeared in the litter microbiota only after chicken placement. Because E. coli is common in the gut microbiota of chickens during the first week of life (11), E. coli in the litter must have been dominant of chicken origin. Although fecal material is the most important factor in changing bedding into the litter, there was also other chicken debris present in the bedding along with their microbiota. Feathers or exfoliated skin epithelial cells released into the environment may act as sources of lactobacilli, saprophytic staphylococci, or Corynebacterium, common in human skin microbiota (20). Staphylococci are common also in the air of chicken barns (21, 22), indicating microbiota recycling between chickens and litter.

Litter microbiota developed over time. This has been reported earlier but previous studies monitored litter microbiota only up to 6 weeks of flock age typical for broiler production. The fact that we extended the monitoring period using material from broiler breeder flocks does not exclude comparison with earlier studies in broilers. These studies mentioned Staphylococci, lactobacilli, and Corynebacterium as common in a litter (5, 8–10). Wang et al. (5) also reported that Yaniella, Staphylococcus, Brevibacterium, and Salinicoccus were more predominant in reused litter than in fresh litter. These observations match with our conclusions on the initial appearance of Staphylococcus saprophyticus, Corynebacterium stantonii (casei), Jeotgallicoccus, and Lactobacillales, including genera Aerococcus, Weissella, and Lactobacillus during the first 2 months followed by Brevibacterium, Brachybacterium, Salinicoccus, Dietzia, Yaniella, Salinicoccus, and Staphylococcus lentus in months 3 and 4. Similar litter microbiota was reported by all authors regardless of the slightly different production technology or initial bedding material used in all different studies, in agreement with the lower importance of initial bedding material for litter microbiota development proposed earlier (1).

How safe is litter microbiota for chickens or humans and what are the most likely sources of litter microbiota? Although it is impossible to conclude on the safety in general, at least some of the litter microbiota members are common in human fermented food products, especially the bacterial species found in months 1 and 2 litter. Staphylococcus saprophyticus, Staphylococcus xylosus, Staphylococcus piscifermentans, Weissella jogaejeotgali, or Corynebacterium casei are common in fermented seafood, sea fish, shrimp, soya sauce or ripening cheese (23–29). Some of them have been reported as uric acid utilisers (30) or histamine degraders (31) or reduced horizontal transfer of antibiotic resistance genes (32). These species can be therefore considered potential microbial products affecting litter microbiota development. Species characteristic for month 3 and 4 litter microbiota were detected in saline soil, sediments, or waste. This is true for Brevibacterium iodinum (33, 34), Brevibacterium metallicus (35), Salinicoccus kunmingensis (36), Staphylococcus lentus (37), Pedobacter lusitanus (38), Salinicoccus salitudinis (39), or Oceanisphaera sediminis (40). Their appearance in an older litter is therefore a consequence of the formation of litter cake with increased osmolarity.

All the species common in the litter can be introduced to the barns by human personnel or wild animals. Brachybacterium, Aerococcus, Brevibacterium, Jeotgallicoccus, Weissella, Facklamia, Oceanobacillus, or Yaniella were detected in farm air dust indicating one common way of their introduction from the external environment (2, 41, 42). In addition, houseflies from the farm environment transferred Corynebacterium, Lactobacillus, Staphylococcus, or Weissella on their surface showing the second common way of their introduction to barns from the external environment (43–45).

When testing to what extent litter microbiota may act as a reservoir of antibiotic resistance for gut microbiota, we concluded that for tetQ, tetW, and tet(44), this was not likely because these genes were more abundant in cecal microbiota than in litter samples. tetO was similarly distributed in the cecal and litter microbiota confirming its broad distribution among different bacterial species. bla_TEM positively correlated with E. coli presents both in the cecum and in the litter, thus confirming the conclusions in a previous report (7). tetA(P) exhibited the highest correlation with Romboutsia timonensis. Although PCR quantification indicated a more frequent distribution of this gene among litter microbiota, Romboutsia is common in the small intestine (46, 47). Thus, the comparison of tetA(P) abundance in a litter with its abundance in cecal microbiota may be inappropriate. If tetA(P) abundance in the litter microbiota was compared with its distribution in ileal microbiota, this gene may be more common in ileal microbiota due to the association with Romboutsia.

In conclusion, we analyzed and compared the microbiota from the bedding material, developing litter, and chicken cecum. We found that these 3 types of samples were populated by different bacterial species which did not overlap extensively. A time-dependent development of litter microbiota was observed. This, together with the positive experience of reusing litter and the appearance of some of these species in fermented food, may open the question of whether certain saprophytic Staphylococci, Corynebacteria, or members of order Lactobacillales may be considered biological cleaning products to actively suppress pathogen multiplication in the litter and/or degrade chicken metabolic by-products.

MATERIALS AND METHODS

Sample collection.

The development of litter microbiota was determined in 6 different flocks from 2 different farms at 3 different time points. The first two flocks of broiler breeder chickens kept in 2 different barns of farm A were monitored from October 2019 until February 2020. An additional two flocks of broiler breeder chickens that were kept in two different barns of the same farm A were followed from April to July 2020. Finally, we monitored bedding and litter microbiota in two broiler flocks kept in two different barns of farm B in August 2020 (Fig. 6). In the latter case, we also sampled the bedding before chicken placement to check for bedding microbiota unaffected by the deposition of chicken fecal material. The two farms represent commercial production settings, belonged to two different owners, and bought 1-day-old chicks from different sources. Farm B was in the same area as farm A, and each of them was within 20 km distance from Brno, Czech Republic. Chicken cecal microbiota was determined in 5 chicks from each of the 6 monitored flocks after placement and at the end of the rearing period. The euthanasia of chickens was performed according to the current Czech legislation (Animal Protection and Welfare Act No. 246/1992) and was approved by the Ethics Committee of the Veterinary Research Institute followed by the Committee for Animal Welfare of the Ministry of Agriculture of the Czech Republic (permit number MZe1922 approved on January 15, 2018).

FIG 6.

Experimental design. The development of litter microbiota was determined in 6 different flocks from 2 different farms at 3 different time points. The broiler breeder chickens were monitored in farm A. Bedding and litter microbiota was monitored in broiler flocks on farm B.

Litter samples were collected weekly, although a strictly weekly sampling plan could not be accomplished for logistic reasons. This meant that for flock 1, we collected 9 litter samples, and for flock 2, 13 litter samples were available. In the repeated sampling, 15 samples were collected for flock 3 and an additional 15 litter samples were collected for flock 4. In the two broiler flocks, we collected 5 straw bedding samples before chicken placement and 45 litter samples.

Altogether we collected 162 samples, of these 5 were bedding samples, 97 were litter samples and 60 were cecal samples. The exact time points when the samples were collected are provided in Table S1 in Supplemental File 1. After preliminary analyses, the samples were grouped irrespective of farm origin according to flock age in months, i.e., 1- to 4-month-old litter samples and cecal samples from 1- and 4-month-old chickens. Bedding samples collected before chicken placement on a farm were treated as month 0 samples.

DNA purification and 16S rRNA sequencing.

All samples were stored at −20°C until further processing. Before DNA purification, the samples were homogenized in a MagNALyzer (Roche), and the DNA was extracted using a QIAamp DNA Stool Minikit according to the manufacturer’s instructions (Qiagen, Germany). The DNA concentration was determined spectrophotometrically, and samples diluted to 5 ng/mL were used as a template DNA in PCR with forward primer 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG-MID-GT-CCTACGGGNGGCWGCAG-3′ and reverse primer 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG-MID-GT-GACTACHVGGGTATCTAATCC-3′ (48, 49). The sequences in italics served for index ligation whereas the underlined sequences allowed for amplification over the V3/V4 region of 16S rRNA genes. Molecular identifier (MIDs) represent different sequences of 5, 6, 7, or 9 bp in length used for sample differentiation. PCR amplification was performed using a KAPA HiFi Hot Start Ready Mix kit (Kapa Biosystems), and the resulting PCR products were purified using AMPure beads. In the next step, the PCR product concentration was determined with a spectrophotometer before the DNA was diluted to 100 ng/μL, and groups of 14 PCR products with different MID sequences were indexed with the same index from the Nextera XT Index kit following the manufacturer’s instructions (Illumina). The next set of 14 PCR products with different MID sequences was indexed with the next index from the Nextera XT Index kit, thus allowing an increase in the number of samples analyzed in a single sequencing run. Before sequencing, the concentration of differently indexed samples was determined using a KAPA Library Quantification Complete kit (Kapa Biosystems). All indexed samples were diluted to 4 ng/μL and 20 pM phiX DNA was added to a final concentration of 5% (vol/vol). Sequencing was performed using MiSeq reagent kit v3 (600 cycles) and MiSeq apparatus according to the manufacturer’s instructions (Illumina). Raw sequencing data were deposited in the GenBank under accession number PRJNA810953.

Microbiota analysis was performed with QIIME 2 (50). Raw sequence data were demultiplexed and quality filtered, and sequencing primers were clipped using Je (51) and fastp (52). The resulting sequences were denoised with DADA2 (53). Taxonomy was assigned to ASVs using the q2-feature-classifier (54) classify-sklearn naïve Bayes taxonomy classifier against the Silva 138 99% ASVs full-length reference sequences (55). All the software tools were used with default settings.

Quantification of selected antibiotic resistance genes.

DNA from all samples was used as a template in quantitative real-time PCR. The following target genes were quantified: bla_TEM as an indicator of Enterobacteriaceae, tetQ as an indicator of Bacteroidetes, tetA(P) as an indicator of pathogenic Clostridia, tetO as ubiquitously distributed gene, and tetW and tet(44) as indicators of Lachnospiraceae and Ruminococcaceae (13, 14).

PCR was performed in 3 μL volumes in 384-well microplates using QuantiTect SYBR Green PCR Master mix (Qiagen). Dispensing of the PCR master mix, primers, water, and DNA was performed using a Nanodrop pipetting station (Innovadyne). PCR and signal detection were performed with a LightCycler II (Roche) with an initial denaturation at 95°C for 15 min followed by 40 cycles of denaturation at 95°C for 20 s, primer annealing at 60°C for 30 s and extension at 72°C for 30 s. Each sample was subjected to real-time PCR in triplicate and the mean values of the triplicates were used for subsequent analysis. Amplification of the 16S rRNA gene using domain Eubacteria-specific primers was used as a reference to determine the total amount of bacterial DNA in each sample. Threshold cycle (Ct) values of genes of interest were subtracted from the Ct value of bacterial 16S rRNA gene amplification (ΔCt) and the relative abundance of each gene of interest was finally calculated as 2^-ΔCt. Primers used for the amplification are listed in Table 2.

TABLE 2.

List of primers used for the quantification of antibiotic resistance genes

Target gene	Forward primer 5′–3′	Reverse primer 5′–3′
blaTEM	GATGGTAAGCCCTCCCGTAT	GGCACCTATCTCAGCGATCT
tetO	ACGGAAAGTTTATTGTATACC	TGGCGTATCTATAATGTTGAC
tetQ	AGAATCTGCTGTTTGCCAGTG	CGGAGTGTCAATGATATTGCA
tetW	AGCGACAGCGTGAGGTTAAA	AAGTTGCGTAAGAGCGTCCA
tet(44)	CGAAAGCAAAGTTTCACTCGGT	AAGCGAAAATCCGAGGGAGT
tetA(P)	AGTTGCAGATGTGTACAGTCG	CTTCCGCAATCCAAGCTTCA
Eubacterial 16S rRNA	TCCTACGGGAGGCAGCAG	CGTATTACCGCGGCTGCT

Statistical analysis.

PERMANOVA (R-project, package vegan, Adonis function followed by pairwise comparisons) using Bray-Curtiss matrix distances was used to determine groups significantly differing in microbial composition. LEfSe (linear discriminant analysis effect size) was used to determine taxa which most likely explained the differences between compared groups (56). Spearman’s correlation was used to calculate correlations between the abundance of a particular family in chicken gut microbiota and the abundance of selected resistance genes. Computed correlations are presented as a heat map. LEfSe and correlation analysis was performed with the top 202 ASVs which were selected as forming at least 0.25% of the total microbiota in at least one of 162 sequenced samples. Sequences of these 202 ASVs were manually annotated using GenBank rRNA/ITS database and such taxonomic designation is used in Fig. 2 and 5. This reduction was done to operate with the common ASVs that likely represented real bacterial species.