Longitudinal Changes in Campylobacter and the Litter Microbiome throughout the Broiler Production Cycle

Abstract

Broiler chickens are an important source of Campylobacter to humans and become colonized on the farm, but the role of the litter in the ecology of Campylobacter is still not clear. The aim of this study was to examine the relationship between Campylobacter and the changes in the litter microbiome throughout the broiler production cycle. Twenty-six commercial broiler flocks representing two production types (small and big broilers) were followed from 1 to 2 weeks after placement to the end of the production cycle. Composite litter samples from the broiler chicken house were collected weekly. Litter DNA was extracted and used for Campylobacter jejuni and Campylobacter coli qPCR as well as for 16S rRNA gene V4 region sequencing. Campylobacter jejuni concentration in litter significantly differed by production type and flock age. Campylobacter jejuni concentration in litter from big broilers was 2.4 log₁₀ units higher, on average, than that of small broilers at 3 weeks of age. Sixteen amplicon sequence variants (ASVs) differentially abundant over time were detected in both production types. A negative correlation of Campylobacter with Bogoriella and Pseudogracilibacillus was observed in the litter microbiome network at 6 weeks of flock age. Dynamic Bayesian networks provided evidence of negative associations between Campylobacter and two bacterial genera, Ornithinibacillus and Oceanobacillus, at 2 and 4 weeks of flock age, respectively. In conclusion, dynamic associations between Campylobacter and the litter microbiome were observed during grow-out, suggesting a potential role of the litter microbiome in the ecology of Campylobacter colonization and persistence on farm. IMPORTANCE This study interrogated the longitudinal association between Campylobacter and broiler litter microbiome in commercial broiler flocks. The results of this investigation highlighted differences in Campylobacter dynamics in the litter throughout the broiler production cycle and between small and big broilers. Besides documenting the changing nature of the microbial networks in broiler litter during grow-out, we detected bacterial genera (Oceanobacillus and Ornithinibacillus) negatively associated with Campylobacter abundance and concentration in litter via the Bayesian network framework. These bacteria should be investigated as possible antagonists to Campylobacter colonization of the broiler environment.

Keywords: Campylobacter; broiler chickens; cohort; litter; microbiome; network.

FIG 1

Campylobacter concentration (via qPCR) in litter from broiler chickens. (A) Campylobacter concentration in litter (gray circles), combining C. jejuni and C. coli concentrations, by flock age. A LOWESS smoothed line and a fitted line from Tobit regression are shown in red and blue, respectively. (B) C. jejuni (red triangles) and C. coli (blue squares) concentrations in litter by flock age. LOWESS smoothed lines and fitted lines from Tobit regression are shown as dashed and dash-dotted lines, respectively, and their color matches the corresponding Campylobacter species (red for C. jejuni and blue for C. coli). Jittering was added to both graphs to improve the visualization of the data points.

FIG 2

Fitted Campylobacter jejuni dynamics in litter by production type and flock age. Vertical bars correspond to 95% confidence intervals.

FIG 3

Alpha diversity indices (fitted values) in litter microbiome by production type and flock age. Vertical bars correspond to 95% confidence intervals.

FIG 4

Beta diversity in litter microbiome by flock age. PCoA: Principal coordinate analysis. Three dissimilarity distances were used: Aitchison, Bray-Curtis, and Jaccard.

FIG 5

Subgraphs from litter microbial networks including Campylobacter genus. Microbial networks were built from litter microbiome throughout the production cycle (one through 9 weeks of flock age) using data from small and big broilers together. Networks from weeks one, four, seven, and eight did not include Campylobacter genus in their components. The line connecting two bacterial genera represents a significant correlation between them (green: positive correlation, red: negative correlation).

FIG 6

Dynamic mixed Bayesian microbiome network in broiler litter. The abundance of bacterial genera was centered log ratio transformed and C. jejuni and C. coli genome equivalents were log₁₀ transformed. Production type, location, farm ID, house, house section, flock order, and sequencing batch, were included as potential confounders in the network. Directed connection (edge or arrow) width is proportional to the corresponding bootstrap support. Connection direction was inferred from the data using Hill-climbing algorithm and Bayesian Information Criterion (BIC) score. Only data from two to 5 weeks of flock age was included. Child variables (nodes) displayed in Table 2 are zoomed in.

FIG 7

Dynamic discrete Bayesian microbiome network in broiler litter. The abundance of bacterial genera was centered log ratio transformed and C. jejuni and C. coli genome equivalents were log₁₀ transformed. Both types of variables were discretized as binary variables (presence/absence). Directed connection (edge or arrow) width is proportional to the corresponding bootstrap support. Connection direction was inferred from the data using Hill-climbing algorithm and Bayesian Information Criterion (BIC) score. Only data from 2 to 5 weeks of flock age was included. Child variables (nodes) displayed in Table 3 are zoomed in.