The Use of Disinfectant in Barn Cleaning Alters Microbial Composition and Increases Carriage of Campylobacter jejuni in Broiler Chickens

ABSTRACT

To maintain food safety and flock health in broiler chicken production, biosecurity approaches to keep chicken barns free of pathogens are important. Canadian broiler chicken producers must deep clean their barns with chemical disinfectants at least once annually (full disinfection [FD]) and may wash with water (water wash [WW]) throughout the year. However, many producers use FD after each flock, assuming a greater efficacy of more stringent cleaning protocols, although little information is known regarding how these two cleaning practices affect pathogen population and gut microbiota. In the present study, a crossover experiment over four production cycles was conducted in seven commercial chicken barns to compare WW and FD. We evaluated the effects of barn cleaning methods on commercial broiler performance, cecal microbiota composition, Campylobacter and Salmonella occurrence, and Campylobacter jejuni and Clostridium perfringens abundance, as well as on short-chain fatty acid (SCFA) concentrations in the month-old broiler gut. The 30-day body weight and mortality rate were not affected by the barn cleaning methods. The WW resulted in a modest but significant effect on the structure of broiler cecal microbiota (weighted-UniFrac; adonis P = 0.05, and unweighted-UniFrac; adonis P = 0.01), with notable reductions in C. jejuni occurrence and abundance. In addition, the WW group had increased cecal acetate, butyrate, and total SCFA concentrations, which were negatively correlated with C. jejuni abundance. Our results suggest that WW may result in enhanced activity of the gut microbiota and reduced zoonotic transmission of C. jejuni in broiler production relative to FD in the absence of a disease challenge.

IMPORTANCE We compared the effects of barn FD and WW methods on gut microbial community structures and pathogen prevalence of broiler chickens in a nonchallenging commercial production setting. The results revealed that barn cleaning methods had little impact on the 30-day body weight and mortality rate of broiler chickens. In addition, the FD treatment had a subtle but significant effect on the broiler cecal microbiota with increased abundances of Campylobacter and decreased SCFA concentrations, which would support the adoption of WW as a standard practice. Thus, compared to FD, WW can be beneficial to broiler chicken production by inhibiting zoonotic pathogen colonization in the chicken gut with reduced cost and labor of cleaning.

INTRODUCTION

In broiler chicken production, biosecurity measures are important to maintain flock health and food safety. Regulations of the current on-farm-food safety assurance program require Canadian broiler chicken producers to clean barns with disinfectants at least annually (1). Barn cleanouts within the year using water washing (WW) can be done without disinfectants; however, in practice many producers continue to perform full disinfection (FD) using chemical disinfectants after each flock. Using chemical disinfectants removes a high proportion of microbes (2) but may also reduce the transmission of beneficial microbes between flocks. This could lead to a potential loss in the microbes that can outcompete pathogens in the environment. It may also result in selection of disinfectant-resistant pathogens that further increases the risk of pathogen contamination of animal products (3). To date, little information is available regarding how these cleaning measures affect chicken health and zoonotic pathogen colonization. Furthermore, emerging evidence has shown the importance of broiler commensal microbial community to nutrient metabolism, feed efficiency (4–6), host resistance to pathogens (7), and immune system development (8, 9). In chicken production, the establishment of a symbiotic microbiota has been shown to increase nutrient utilization (10) and prevent disease development (11). Therefore, establishing healthy host-microbe interactions early in production may provide a possible solution to help maintain or even enhance broiler performance in an environment free from antimicrobial growth promoters. Broiler gut microbiota assembly and development is significantly influenced by the initial environment to which they are exposed (12). Thus, placing newly hatched broilers in an environment with microbiota from a previous healthy flock may advance the development of a commensal microbiota in broiler chickens.

Previous work has shown that exposure to reused broiler litter altered early-life gut microbiota and increased infection resistance to pathogens in broiler chickens (13–15). Reused litter induced changes in the broiler gut microbiota of chicks over the first 2 weeks of life, resulting in an increased predominance of Clostridiales in the gut (16). More recently, the application of reused litter was reported to increase the predominance of some potentially beneficial bacteria, such as Faecalibacterium, a short-chain fatty acid (SCFA) producer whose increased abundance in young broiler ceca continued as the chicken matured (17). Commensal microbes and SCFAs are important in maintaining gut homeostasis. For example, butyrate increases intestinal epithelial oxygen consumption, which helps to maintain an anaerobic environment (18). SCFAs also modulate host immune responses by suppressing proinflammatory cytokine expression to achieve homeostasis (19).

Both poultry and zoonotic pathogens have long been a major concern to the poultry industry. The poultry gut microbiota plays an important role in pathogen exclusion. For example, commensal members of the chicken gut microbiota inhibited Salmonella enterica colonization by competitive exclusion (15, 20). Colonizing broiler chicks with a diverse set of adult chicken commensal bacterial isolates reduced Campylobacter jejuni colonization (21). In addition to the potential impact on human health by zoonotic diseases, the economic impact of poultry diseases is a major concern of chicken producers. Bacterial pathogens such as Clostridium perfringens can cause necrotic enteritis, resulting in reduced growth and feed efficiency, and in severe cases, increased mortality (22). To date, in-feed antimicrobial additives are used in broiler farming to control necrotic enteritis. Efforts have been made to find effective alternatives to control necrotic enteritis. Concerns over the spread of antimicrobial resistance from agriculture have led to legislation and consumer pressure to reduce antibiotic use in animal production globally. Previous research has demonstrated that using a cocktail of probiotics reduced the level of cecal C. perfringens and associated intestinal lesions, which showed the potential to be an alternative to antibiotics (23). However, antibiotics remain the most effective and widely used approach to keep the prevalence of C. perfringens infections low in the poultry industry (24).

To date, limited information is available on how WW and FD affect pathogen prevalence, gut microbial communities, nutrient metabolism, and host responses in chickens. Here, a crossover experiment was designed using seven similar commercial chicken barns to compare WW and FD over the course of four production cycles at two locations. We evaluated the effects of barn cleaning method on commercial broiler intestinal microbiota, the occurrence of Campylobacter and Salmonella, and the abundance of C. jejuni and C. perfringens, as well as on the SCFA profile, in 30-day-old broiler ceca.

RESULTS

Chicken 32-day body weight and mortality were not affected by the cleaning methods.

The 32-day average body weight in the WW group was comparable to the FD group (Fig. 1, left, P = 0.22). In addition, no difference in 30-day mortality was observed between the two barn cleaning treatments (Fig. 1, right, P = 0.91), suggesting that the cleaning method had a minimal impact on the flock performance.

FIG 1.

Broiler chicken flock performance at 32 days of age. (Left) Flock mean body weights at day 30 of 1,782 ± 30.09 and 1,780 ± 20.59 g for FD and WW, respectively (n = 12 flocks/treatment, means ± the SEM; FD, full disinfection; WW, water wash); (right) flock mean mortality rates at day 30 of 0.061 ± 0.0074 and 0.059 ± 0.0071 for FD and WW, respectively (n = 14 flocks/treatment, means ± the SEM; FD, full disinfection; WW, water wash).

Chicken barn FD resulted in increased Campylobacter occurrence in 30-day-old chicken ceca.

Salmonella was not detected through enrichment in any of the samples collected through the study, therefore, the impact of treatment on Salmonella shedding could not be assessed. To evaluate the effect of cleaning method on the pathogen occurrence, an occurrence scoring method was used. In the Campylobacter enrichment assay, no Campylobacter was detected from the litter samples, feeding pan and drinker swabs, or the shoe-cover samples before broiler placement. At the broiler placement day, the Campylobacter occurrence scores were not different between the two treatments (see Fig. S1 in the supplemental material). However, at day 30, the WW group exhibited a significantly lower Campylobacter occurrence score compared to the FD group (P < 0.05) (Fig. 2). Campylobacter was detected in at least one sample from each FD flock. Therefore, the WW rearing environment reduced the occurrence of Campylobacter colonization in 30-day-old broiler ceca.

FIG 2.

Broiler chicken cecal Campylobacter occurrence score at 30 days of age. The results show the mean flock score ± the SEM (n = 14/treatment; *, P < 0.05). Campylobacter occurrence score = number of pathogen positive broilers/total number of broilers sampled per barn. WW, water wash; FD, full disinfection.

Chicken barn FD increased C. jejuni abundance.

The quantification of cecal microbial load using quantitative PCR (qPCR) showed that the WW group did not differ from the FD group in cecal C. perfringens load (5.81 and 6.06 log₁₀ copies/g of netB for the WW and FD groups, respectively; P = 0.20) (Fig. 3a). Consistent with the Campylobacter enrichment assay results, the WW group exhibited ∼0.9-log₁₀ lower hipO copy numbers compared to the FD group (Fig. 3b, 8.13 and 9.03 log₁₀ copies/g of hipO for the WW and FD group, respectively; P < 0.05), indicating that the decreased sanitation stringency reduced C. jejuni colonization in the mature chicken ceca. Furthermore, the barn cleaning method did not affect the total bacterial load in the chicken ceca (Fig. 3c; P = 0.15).

FIG 3.

Broiler chicken (30 days of age) cecal qPCR targeting the C. perfringens netB gene (a), the C. jejuni hipO gene (b), and the bacterial 16S rRNA gene (c). The results show the average copy number of each target gene (mean ± the SEM, n = 70/treatment; **, P < 0.01; n.s., P > 0.05). WW, water wash; FD, full disinfection.

Barn cleaning methods had subtle impacts on the 30-day-old chicken microbiome.

On average, 42,943.3 ± 2,757.8 (mean ± the standard errors of the mean [SEM]) reads per cecal sample were generated and processed by QIIME2 pipeline, resulting in a total of 3,845 ASVs. For litter samples, an average of 7,847.8 ± 1,209.2 (mean ± SEM) reads per sample were generated and processed, resulting in 1,780 amplicon sequence variant (ASVs). When focusing on the effect of different cleaning methods, the cecal microbiomes of broilers from the WW and FD groups had comparable richness and evenness, as indicated by alpha-diversity indices (Chao1, P = 0.71; Shannon, P = 0.25) (Fig. 4). Beta-diversity analyses based on both weighted and unweighted UniFrac matrices suggested that cecal microbial communities in the WW group differed from the FD group (adonis P = 0.05, R² = 0.012 and adonis P < 0.01, R² = 0.013 for weighted UniFrac and unweighted UniFrac matrices, respectively) (Fig. 5). Differences in abundances of two bacterial taxa were also suggested by ANCOM (Fig. 6). The genus Helicobacter was more predominant in the WW group (W = 85), whereas the family Bacillaceae was more predominant in the FD group (W = 66). These results suggested that the barn cleaning treatments influenced the relative abundance of two bacterial taxa and in turn led to a modest but significant impact on the overall structure of the chicken gut microbiota. However, the cleaning method did not lead to changes in microbial structures of the litter samples (see Fig. S2 and S3).

FIG 4.

Alpha diversity of broiler chicken cecal microbiome at 30 days of age. Box plots show the alpha diversity in samples determined using the Chao1 index and Shannon index (n = 70/treatment, n.s., P > 0.05). WW, water wash; FD, full disinfection.

FIG 5.

Principal coordinate analysis plots based on weighted and unweighted UniFrac distance matrices. Barn cleaning treatments had modest but significant effects on microbial community structure in the chicken ceca at 30 days of age (n = 70/treatment). WW, water wash; FD, full disinfection.

FIG 6.

Relative abundances of genus Helicobacter and family Bacillaceae. The bar plots show the relative abundances of taxa of interests of each treatment with individual values (n = 70/treatment, means ± the SEM). g__, genus; f__, family.

Cecal SCFA profile differed by cleaning methods.

Cecal contents were subjected to gas chromatography to measure the cecal SCFA concentration. The WW group showed significantly greater total SCFAs than the FD group (P < 0.01) (Fig. 7a). Specifically, acetate (Fig. 7b), propionate (Fig. 7c), and butyrate (Fig. 7d) concentrations in the WW group were higher than those in the FD group. A trend for higher valerate concentration (Fig. 7e; P = 0.06) was also observed in WW broilers. Spearman correlation between SCFA concentration and bacterial relative abundance suggested a series of microbes that are correlated with the altered SCFA profile between treatments (Fig. 8). Total SCFAs and acetate concentration were negatively correlated with the genus Campylobacter and members from the orders RF32 and YS2 (P < 0.05). On the other hand, an unclassified genus belonging to the order of Clostridiales was positively associated with total SCFA, acetate, and butyrate concentrations (P < 0.05). Propionate concentration was negatively associated with the genus Lachnospira (P < 0.01) and an unclassified genus of the family Enterobacteriaceae (P < 0.05), whereas it was positively associated with the genus Odoribacter (P < 0.05) and an undetermined genus of the family Clostridiaceae (P < 0.05).

FIG 7.

Cecal SCFA production in broiler chickens at 30 days of age. The results show the averages total SCFA (a), acetate (b), propionate (c), butyrate (d), and valerate (e) concentrations (means ± the SEM, n = 20/treatment; *, P < 0.05; **, P < 0.01; n.s., P > 0.05). FD, full disinfection; WW, water wash; Conc., concentration.

FIG 8.

Heatmap showing Spearman correlations between cecal bacterial abundance and SCFA concentrations in broiler chickens at 30 days of age. *, P < 0.05; **, P < 0.01; Conc., concentration.

DISCUSSION

This study was the first to characterize the impact of FD and WW on chicken gut microbiota in a commercial setting. Previously, de Castro Burbarelli et al. examined the effects of poultry barn cleaning using neutral detergent versus a protocol using acidic and alkaline detergent with chemical disinfectants (2). In accordance with these results, we found that compared to FD, the WW method was not associated with compromised flock mean body weight or increased mortality rate at day 32 in production. Unfortunately, we were not able to collect feed consumption data and are therefore not able to comment on any potential impacts on feed conversion.

Although limited research has explored how barn cleaning practices affect the development of chicken intestinal microbial communities, especially in the context of commercial production, efforts have been made to study the effect of reused litter on chicken gut microbiota. Some laboratory-scale research suggested that reused litter mainly influences broiler gut microbiota at early ages. Cressman et al. reported that compared to the reused litter group, broilers provided fresh litter had a greater bacterial alpha diversity in ceca at 7 days of age (16). However, no treatment effects on the gut microbiota were observed in the later time points. In addition, broilers reared on fresh litter were colonized by microbes identified in fresh litter, including Lactobacillus, unclassified Lachnospiraceae, and Enterococcus, whereas broilers reared on reused litter were colonized by typical poultry intestinal bacteria, such as members of the order Clostridiales (16). Similarly, Wang et al. reported that at both day 10 and day 35 of age, the broiler gut microbiota was altered by the reused litter treatment with increased predominance of Faecalibacterium prausnitzii in ceca (17). In the present study, no treatment effect on the alpha diversity of the 30-day-old broiler gut microbiota was observed, and only a modest effect was shown on the beta diversity between FD and WW treatments. This may be explained by the fact that fresh litter, which acts as a physical barrier, was placed in both WW and FD barns. Here, we did not observe differences of the litter microbiota between treatments. In addition, reused litter may provide a more functional ecological niche compared clean litter. Generally, reused chicken litter is a mixture of bedding material and excreta, which offers more surfaces and available nutrients for microbes to attach and survive on. In addition, since chickens are coprophagic, the consumption of litter material likely increases the opportunity for successful microbial transmission from one flock to the next (25). In the present study, only five broilers per flock were selected for 16S rRNA gene amplicon sequencing and Campylobacter and Salmonella enrichment. While a larger sample size would strengthen the conclusions, previous microbiome studies report that individuals housed together, particularly coprophagic animals, show less variation in the intestinal microbiota with strong cohousing effects (26, 27). A clear flock effect was also observed in the present study (data not shown). Rather than sampling more broilers from each barn, we chose to sample more barns. In addition, the crossover design of the animal trial can also help eliminate bias brought by management and/or housing facilities.

Helicobacter was found to be less abundant in the FD group at day 30 compared to the WW group. Interestingly, Helicobacter is a genus identified as a disappearing member of the human gut microbiome and may also be associated with increased use of disinfectants (28). In avian species, members in genus Helicobacter have been detected in wild birds (29). Studies on the relationship between Helicobacter spp. and a chicken host are highly variable. Some members of genus Helicobacter are considered opportunistic pathogens in chickens. For example, Helicobacter pullorum infection was found to cause mild lesion in the chicken ceca (30). However, the effects of Helicobacter on host health can vary between different bacterial strains within species (31). Yin et al. reported that Helicobacter abundance increased in response to α-amylase, amylopectase, and glucoamylase supplementation in a corn-based diet and was associated with increased starch digestibility and higher mature bodyweight (32). In the present study, Helicobacter positively correlated with branched-chain fatty acids (BCFAs) isobutyrate and isovalerate (Fig. 7). BCFAs are often used as indicators of protein catabolism (33). Currently, the direct relationship between BCFAs and their impact on host health is still unclear (34). It is reported that BCFAs modulate adipocyte lipid and glucose metabolism and contribute to increased insulin sensitivity (35). With 16S rRNA gene amplicon sequencing, it is difficult to discriminate bacteria to the species or strain level. Therefore, our identification of Helicobacter as the genus of increased predominance in the WW group needs to be further studied. In addition, information on metabolic functionality is also warranted to understand the role of Helicobacter in the chicken gut.

Interestingly the FD broilers showed increases in the relative abundance of the family Bacillaceae. Members of the Bacillaceae family, such as Bacillus are Gram-positive, rod-shaped bacteria that can form endospores to survive in harsh physical and chemical environments (36). It is possible that the chemical conditions given by FD treatment provided a selective pressure that led to the increased level of Bacillaceae. Some members in this order have the ability to produce antimicrobial peptides and are recognized as potential beneficial bacteria (37). Some Bacillus species, such as Bacillus licheniformis and Bacillus subtilis, have been commercially added into poultry feed as probiotics (38). However, not all Bacillus species are beneficial. Bacillus cereus is a foodborne pathogen that causes diarrhea (39). Recently, the prevalence of the B. cereus group made up 50% of Bacillus isolates from retail chicken products (40). In addition, some non-B. cereus species were found to carry virulence genes and exhibited the same phenotypic virulence characteristics as B. cereus (41). Furthermore, tests of antimicrobial resistance have identified multidrug-resistant isolates regardless of virulence factors, indicating that further evaluation of the impact of Bacillus on food safety and public health is needed (40). In this sense, the effect of barn disinfection on increasing Bacillaceae may need to be carefully assessed.

The reduced Campylobacter load in the present study with WW treatment is consistent with previous studies showing that reused litter has the potential to reduce gut pathogen abundance in broiler chickens. It has been reported that reused litter reduced the colonization of Salmonella enterica serovar Typhimurium and Salmonella enterica serovar Enteritidis in infected chickens (13, 42). In the present study, although Salmonella was not detected in any of samples by the enrichment assay, our results supported that the WW treatment did not increase Salmonella occurrence in the chicken intestine. Changing from FD to WW for within-year washing would be associated with lower labor and material costs for cleaning. To date, limited data are available regarding the effects of cleaning treatments and disinfectants on Campylobacter occurrence and abundance in the chicken intestine. de Castro Burbarelli et al. reported a trend that Campylobacter was more frequently detected in the intestine of the stronger disinfection group (2), although the study was conducted in a controlled research setting. Furthermore, it has been shown that colonization of young chicks with bacterial cocktails of mature chicken commensal isolates reduced colonization by C. jejuni (43). Therefore, the FD treatment in the present study may have eliminated some microbes from the previous flock which can potentially compete with Campylobacter.

While the effects of cleaning method on the gut microbial composition were relatively modest, changes in the concentrations of microbial metabolites, SCFAs, were observed. SCFAs enhance intestinal integrity as direct energy sources to enterocytes (44). Moreover, complex interactions between SCFAs, gut microbes, and the host immune system have been well documented (45). Briefly, SCFAs from intestinal microbe fermentation are imported into enterocytes and tissues via transporters and paracellular transport. SCFA receptors expressed on enterocytes and immune cells in the lamina propria and mucosal lymphoid tissue can activate signaling pathways to regulate host immune response according to the SCFA concentration to maintain intestinal homeostasis (45). Butyrate signaling through G-protein-coupled receptors can confer anti-inflammatory properties in colonic dendritic cells by downregulating the expression of cytokines and chemokines (19). In the present study, there were negative correlations between the relative abundance of Campylobacter and total SCFA, acetate, and butyrate concentrations in ceca. It has been suggested that SCFAs, especially butyrate, had shown bactericidal effect on Campylobacter in vitro (46). More recently, Awad et al. reported that C. jejuni infection led to reduced acetate and butyrate concentrations in chicken ceca (47). Adding microencapsulated butyrate to feed was also found to reduce Campylobacter colonization in the chicken intestine (48). Together with our results, it is reasonable to suggest that the FD treatment discriminated against beneficial commensals in the gut environment, which could compete with or inhibit Campylobacter by producing SCFAs.

Conclusion.

Compared to FD, we found that WW can be beneficial to broiler chicken production by inhibiting Campylobacter jejuni colonization in the chicken gut with reduced cleaning costs. Further studies examining other barn disinfection practices and testing for other pathogens are warranted to identify the best practices to minimize pathogen load and maintain animal performance.

MATERIALS AND METHODS

Broiler production house and barn cleaning.

A commercial broiler company in AB, Canada provided all the chickens and facilities, including a total of seven similarly engineered single-story, cement-floored production houses at two locations for this study. The broiler facilities were environmentally controlled metal houses with solid sidewall ventilation. Four alternating water and feed lines ran the entire length of each house.

For FD, chicken manure, used litter, and organic matter were completely removed from the chicken house after depopulation followed by a two-step disinfection: (i) All surfaces were thoroughly covered with foam containing 7% sodium hydroxide, 7% 2-(2-2-butoxyethoxy) ethanol, 6% sodium laureth sulfate, 5% sodium N-lauroyl sarcosinate, and 5% tetrasodium EDTA for 60 min, and subsequently rinsed with a high-pressure followed by a low-pressure water wash with the water temperature set at 35°C; (ii) after the broiler house was air-dried, all surfaces were covered with foam containing 10% glutaraldehyde, 10% benzalkonium chloride, and 5% formic acid for 60 min, followed by a high-pressure water rinse. After the two-step disinfection, broiler houses were left to air dry overnight, followed by placement of fresh wood shavings (∼10 to 15 cm deep). For WW, manure and used litter were removed, followed by a low-pressure water rinse with the water temperature set at 35°C of the facility surfaces, air dried, and placement of fresh wood shavings (∼10 to 15 cm deep). The present study was performed on 28 production flocks, and the FD and WW treatments were each applied to 14 production flocks. For each chicken barn, two flocks of each treatment were assigned according to the schedule presented in Table 1.

TABLE 1.

Production cycles, barns, and cleaning treatment schedule

Cycle	Cleaning methoda
	Barn A	Barn B	Barn C	Barn D	Barn E	Barn F	Barn G
1	FD	FD	FD	FD	FD	WW	WW
2	FD	FD	WW	WW	FD	WW	WW
3	WW	WW	WW	WW	WW	FD	FD
4	WW	WW	FD	FD	WW	FD	FD

^aWW, water wash; FD, full disinfection.

Chicken management and sample collection.

Animal use for this experiment was approved by the Animal Care and Use Committee: Livestock of the University of Alberta according to Canadian Council on Animal Care guidelines (49). For each flock, 14,000 Ross 308 broiler chicks were placed at 1 day of age and confined to half of the house, then allowed access to the entire house at 7 days of age. All chickens were fed ad libitum and reared from 1 day of age through processing at about 32 to 35 days of age when the average target live weight of 1.8 kg is reached. Each flock had a placement based on a maximum stocking density of 30 kg/m². Overall mortality rate and body weight of broilers at day 32 were recorded for each barn.

To investigate whether Campylobacter and Salmonella were present after barn cleaning, environmental samples, including litter samples, feeding pan samples, and drinker samples, as well as shoe-cover samples, were collected 3 to 6 h before chicken placement of each flock. For litter samples, ∼50-g subsamples of bedding material were collected from distinct areas in the barn: along water and feeder lines, between water and feed lines, near the wall corners, near the barn entrance, and near the ventilation fans. From each of these areas, litter samples were collected from the surface to the cement floor with sterile gloves and Whirl-Pak sample bags (Whirl-Pak, Madison, WI). Subsamples were sealed and transported to the laboratory on ice. Immediately after arriving the laboratory, samples from the same barn were pooled (resulting in ∼200 g of litter/barn) in a sterile sampling bag containing 250 mL of 0.1% buffered peptone water and homogenized by a Seward Stomacher 400 (Seward, Worthing, West Sussex, United Kingdom) for 1 min.

To collect feeding pan and drinker swabs, Whirl-Pak Speci-Sponge Sampling Bag (Whirl-Pak) were used. For feeding pan swabs, 10 mL of 0.1% buffered peptone water (Oxoid, Basingstoke, Hampshire, UK) was added to the sampling bag to rehydrate the sampling sponge. About 50 cm² of the feeding pan surfaces were carefully swabbed using the rehydrated sampling sponge. For drinker swabs, ∼5 mL of the water coming from the drinking nipple was collected using the dehydrated sponge, followed by swabbing ∼15 cm of the waterline on both sides of the drinking nipple. In each barn, 10 feeding pans and 10 drinkers in multiple locations were randomly selected to subsample. Each sampling bag was used to swab one drinking nipple or one feeding pan. Sterile gloves were changed before each swab. Subsamples were sealed and transported to the laboratory on ice. Immediately after arriving at the laboratory, the 10 drinker/feeding pan subsamples from the same barn were pooled in a sterile stomacher bag aseptically. Buffered peptone water (0.1%) was added to a 1 in 10 dilution. Pooled samples were homogenized for 1 min at 260 using a Seward Stomacher 400 before processing for Campylobacter and Salmonella detection.

To collect shoe-cover samples, sterile DuPont Tyvek shoe covers (Wilmington, DE) were used. Briefly, three layers of sterile shoe-covers were put on at the entrance of the barns before environmental sampling. Shoe-covers were used to walk through the barn following the feeder and water lines. After walking through the whole barn, the outer layers of the shoe-covers were collected using a Whirl-Pak sample bag (Whirl-Pak). Sample bags were sealed and put on ice before arriving at the laboratory. In the laboratory, the shoe-cover sample was transferred to a sterile stomacher bag with 100 mL of 0.1% buffered peptone water, followed by homogenization for 1 min using a Seward Stomacher 400 before processing for Campylobacter and Salmonella detection.

At broiler placement day and day 30, five broilers per flock were euthanized using cervical dislocation for sampling. To ensure representative sampling, the five broilers were each randomly selected from different areas within each barn.

Broiler cecal content collections were conducted aseptically. Briefly, the sampling table was cleaned with 70% ethanol before and between every broiler dissection. Tools and collection tubes were autoclaved and tubes were sealed until samples were added. Approximately 100 mg (at placement) and 300 mg (day 30) of cecal contents were collected and placed in sterile 2-mL Eppendorf tubes containing 1 mL of 0.1% buffered peptone water (Oxoid) for the detection of cecal Campylobacter and Salmonella by enrichment. In addition, ∼500-mg samples of cecal contents were frozen on dry ice immediately until transported to the lab and then stored at −80°C for subsequent DNA extraction.

Campylobacter and Salmonella enrichment.

For genus Campylobacter enrichment and detection, ∼50-mg samples of cecal contents were homogenized with 950 μL of sterile peptone water. Homogenized cecal contents, litter samples, and feeding pan and drinker swabs, as well as shoe-cover samples, were incubated at 37°C overnight, followed by inoculation in Bolton Campylobacter selective broth at a 1:10 ratio (Oxoid) at 42°C for 24 h under microaerobic conditions (5% O₂, 10% CO₂, 85% N₂). Aliquots (100 μL) were serially diluted and spread onto Preston Campylobacter selective agar (Oxoid). For Salmonella enrichment, universal preenrichment broth (Sigma-Aldrich, Oakville, ON, Canada) was used to homogenize cecal contents and environmental samples in 1:10 ratio. All homogenized samples were incubated aerobically at 35°C for 24 h. The cultured broth was transferred to 10 mL of tetrathionate broth (Difco/Becton Dickinson, Sparks, MD) and to 10 mL of selenite cystine broth (Difco). The tetrathionate broth and selenite cystine broth were incubated for 24 h at 42 and 35°C, respectively. After incubation, tetrathionate and selenite cystine broths were streaked onto xylose-lysine-tergitol 4 agar (XLT-4; Difco) and brilliant green sulfa agar (BGS; Difco) plates, respectively. Plates of XLT-4 and BGS agar were incubated at 35°C for 24 h.

The occurrence score for detected pathogens was calculated to evaluate the effect of cleaning method on pathogen occurrence in the chicken ceca. Occurrence score was defined as the number of positive broilers divided by the total number of broilers sampled from the same flock within each barn.

DNA extraction and microbiome analyses.

Total DNA was extracted from homogenized litter samples and cecal contents using the QIAamp Fast DNA Stool minikit (Qiagen, Valencia, CA) with an additional bead-beating step with ∼200 mg of garnet beads at 6.0 m/s for 60 s (FastPrep-24 5G instrument; MP Biomedicals, Inc., Santa Ana, CA). Amplicon libraries were constructed according to the manufacture protocol from Illumina (16S Metagenomic Sequencing Library Preparation) targeting V3-V4 region of the 16S rRNA gene (primers: forward [5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′] and reverse [5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′]). An Illumina MiSeq Platform (2 × 300 cycles; Illumina, Inc., San Diego, CA) was used for a paired-end sequencing run. All sequences were submitted to NCBI Sequence Read Archive under BioProject ID PRJNA767330. The quality of sequencing reads was assessed using FastQC. Sequenced reads were processed using Quantitative Insight into Microbial Ecology2 (QIIME2; 2020.6) (50). The Divisive Amplicon Denoising Algorithm 2 was used to denoise and generate paired-end representative read with truncation lengths of 280-bp forward and 260-bp reverse reads (51). Amplicon sequence variant (ASV) feature table was created based on the denoised results. Qiime2’s q2-feature-classifier was used to assign taxonomy (52) with the pretrained classifier “Greengenes 13_8” (99% identity) (53). Analyses of diversity were done by using the “diversity core-metrics-phylogenetic” command, normalizing to a sampling depth set by the sample with the lowest number of reads (n = 17,309). Chao1 and Shannon diversity indices were calculated by using “diversity alpha-phylogenetic.” The significance of the alpha diversity was determined by using “diversity alpha-group-significance.” Beta diversity was determined in QIIME2 using the unweighted and weighted UniFrac distance metric and a principal coordinate analysis was plotted using phyloseq package in R (version 3.6.1). Permutational multivariate analysis of variance based on the unweighted and weighted UniFrac distance matric was used to determine whether there were significant differences in community structures between treatments (adonis function). Differentiate taxa relative abundance between treatments was determined by the analysis of composition of microbiome in QIIME2 (50).

qPCR.

A qPCR assay was used to quantify C. jejuni and C. perfringens, as well as total cecal bacteria, using hippurate hydrolase (HipO), necrotic enteritis B-like toxin (NetB), and the targeted 16S rRNA gene, respectively (Table 2). Total genomic DNA was extracted from cecal contents as described above. The concentration of DNA was determined by a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA). PerfeCTa SYBR green Supermix (Quantabio, Beverly, MA) was used for qPCR assays which were conducted on an ABI StepOne real-time system (Applied Biosystems, Foster City, CA) following the setup of 95°C for 3 min and 40 cycles of 95°C for 10 s and 60°C for 30 s. To generate targeted gene standards, the 16S rRNA, hipO, and netB genes were amplified by PCR using the primers listed in Table 2. Concentrations of the amplified gene fragments were determined by using a Quant-iT PicoGreen dsDNA assay kit (Invitrogen, Waltham, MA) and used to create a standard curve. Gene copy numbers were determined using the relative standard curve method and normalized to the weight of cecal content used for bacterial DNA extraction.

TABLE 2.

Primers used for qPCR assay of broiler chicken cecal samples collected at 30 days of age

Target gene	Primer		Reference
	Orientation	Sequence (5′–3′)
Total bacterial 16S rRNA gene	Forward	CGGYCCAGACTCCTACGGG	54
	Reverse	TTACCGCGGCTGCTGGCAC
hipO	Forward	TCCAAAATCCTCACTTGCCATT	55
	Reverse	TGCACCAGTGACTATGAATAACGA
netB	Forward	TGATACCGCTTCACATAAAGGTTGG	56
	Reverse	ATAAGTTTCAGGCCATTTCATTTTTCCG

SCFA analysis.

Approximately 30 mg per sample of snap-frozen cecal content was weighed, followed by homogenization with 25% phosphoric acid. Samples were centrifuged at 21,130 × g for 10 min, and the supernatant was collected and filtered using a 0.45-μm-pore size filter. Isocaproic acid (23 μmol/mL) was added at a 1:4 ratio to samples as an internal standard. Samples were analyzed on a Bruker Scion 456 gas chromatography instrument (Bruker, Billerica, MA).

Statistical analyses.

Unless otherwise stated, statistical analyses were conducted using Prism 8 (GraphPad Software, San Diego, CA). Because no location effects were observed for any measurements (i.e., chicken performance and gut microbial structure), all data were analyzed based on treatment across location. Statistically significant differences were determined (P < 0.05) by an unpaired Student t test for parametric data (i.e., analyses of performance, qPCR, and SCFAs). The Kruskal-Wallis test was used to determine the significance of nonparametric data (i.e., microbiome alpha-diversity indices). The Spearman correlation was used to correlate SCFA concentration and bacterial relative abundance. The correlation significance was determined by using the psych package and visualized using the corrplot package in R (version 3.6.1).