Comparative Genomic Characterization of Riemerella anatipestifer Isolates from Chickens, Ducks, and Geese in China Authors: Yunqing Guo, Rongrong Zhang, Tengfei Zhang, Wenting Zhang, Qin Lu, Qiao Hu, Qingping Luo Published: 2026 ABSTRACT Simple SummaryRiemerella anatipestifer causes severe contagious disease in waterfowl and has increasingly been detected in chickens, threatening sustainable poultry production. To investigate host-associated genomic differences, we performed comparative genomic analyses of 158 Chinese field isolates from chicken, duck and goose. The core-genome phylogeny showed no apparent separation according to host, sampling period, or geographic origin. However, virulence and antimicrobial resistance genes exhibited host-associated distribution patterns. Pan-genome analysis identified host-associated differences in genes involved in cell-surface modification and stress responses. Overall, this comparative analysis of chicken-, duck-, and goose-derived isolates provides insight into host-associated genomic differences in R. anatipestifer and may inform future research on disease prevention and control. 1. Introduction Riemerella anatipestifer is a Gram-negative, non-motile, rod-shaped bacterium belonging to the family Flavobacteriaceae. The primary hosts of R. anatipestifer are waterfowl, including ducks and geese, as well as turkeys. Although chickens are considered less susceptible to infection, increasing outbreaks of chicken-derived R. anatipestifer infection have been reported in major poultry-producing regions, including China, Greece, and Poland [1,2]. In waterfowl, R. anatipestifer infection primarily causes septicaemia and infectious serositis and can result in substantial economic losses owing to high incidence and mortality rates among young birds [3]. Chickens infected with R. anatipestifer exhibit clinical signs similar to those seen in infected waterfowl. Furthermore, infection can markedly reduce egg production and hatchability. In chickens, infection has been associated with “jelly-like embryos” and fibrinous lesions such as airsacculitis, arthritis, and oviduct obstruction [4]. Multiple independent duck-to-chicken spillover events have also been identified through ancestral state reconstruction [5]. Despite the increasing detection of chicken-derived R. anatipestifer, the genomic differences among isolates from different poultry hosts remain incompletely characterized. Moreover, R. anatipestifer genomes have been reported to harbor numerous putative antimicrobial resistance (AMR) determinants. Therefore, we aimed to characterize host-associated genomic differences in gene content, putative virulence factors, and AMR determinants among R. anatipestifer isolates from different hosts and time periods. The findings provide genomic data that may facilitate further investigation of the epidemiology and control of R. anatipestifer. 2. Materials and Methods 2.1. Bacterial Strains The genome sequences of 158 R. anatipestifer strains were analyzed, including 40 strains isolated in this study and 118 genomic sequences downloaded from the NCBI database. The 40 strains isolated in this study were obtained from independent farms with diseased commercial broilers and layer farms across Shandong, Hubei, and Liaoning provinces. The 118 R. anatipestifer strains were randomly sampled from all Chinese R. anatipestifer isolates available in NCBI from 2011 to 2025. For isolates from the same region and sampling year, a single strain was randomly selected to avoid duplication, while maintaining coverage of all the available sampling years, regions, and host species. Detailed information on the strains is provided in Table S1. 2.2. Strain Isolation and Genome Sequencing Anal swabs were collected, streaked on tryptic soy agar (TSA) supplemented with 5% newborn calf serum and incubated at 37 °C in 5% CO2 for 48 h. Bacterial cultures were prepared by inoculating single colonies into tryptic soy broth (TSB) supplemented with 5% calf serum and incubating them at 37 °C with shaking at 180 rpm overnight. R. anatipestifer identity was confirmed using 16S rDNA PCR (RA-5: 5′-TCGAGATTGCATCACTTCGCATTG; RA-3: 5′-GCTAGTCTTGAGTATAGTTGA GCT AGC-3′) [2]. PCR was performed using primers with an annealing temperature of 50 °C, yielding a 645 bp amplicon. Genomic DNA was extracted using a Bacterial DNA Kit (Tiangen, Beijing, China) according to the manufacturer’s instructions. Genome sequencing was performed by Shanghai Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). Genomic DNA was fragmented to approximately 400 bp using a Covaris M220 Focused-Acoustic Shearer (Covaris, Woburn, MA, USA). DNA libraries for Illumina paired-end sequencing were constructed from the fragmented DNA utilizing the NEXTFLEX Rapid DNA-Seq Kit (Revvity, Waltham, MA, USA). The final libraries were sequenced using 2 × 150 bp paired-end sequencing on an Illumina NovaSeq 6000 platform (Illumina Inc., San Diego, CA, USA). Raw sequencing reads were preprocessed and quality-filtered using fastp version 0.19.6 (https://github.com/OpenGene/fastp, accessed on 25 August 2026). Raw reads were processed as follows: trim read termini with base Phred quality scores below Q20, filter out reads containing >10% ambiguous N-bases, and discard fragments shorter than 25 bp after adapter clipping and quality trimming. The filtered reads were subsequently assembled de novo using SPAdes version 3.15.5 (http://cab.spbu.ru/software/spades/, accessed on 25 August 2026). Genome coverage calculated based on read mapping reached 100%. The genome coverage of the isolates, calculated via the read-mapping approach, reached 100%, demonstrating full capture of the target genomic sequences. 2.3. Phylogenetic Analysis Coding sequences were uniformly predicted using Prokka (version 1.15.6) with the--noanno, --norrna, and --notrna options, generating standardized protein-coding sequences and GFF/FAA files. The pan-genome was inferred using Panaroo (version 1.6.0) with strict cleaning, gene refinding disabled (--refind-mode off), and a core-gene threshold of 95%. Core genes were aligned using MAFFT (version 7.526). Recombinant regions were identified and masked using Gubbins (version 3.4.3). RAxML-NG (version 2.0.2) was used for construct maximum-likelihood tree construction during Gubbins iterations, with a maximum of five iterations. The final maximum-likelihood phylogeny was inferred using IQ-TREE (version 2.4.0) with ModelFinder and the MFP + ASC model-selection strategy. Branch support was assessed using 1000 ultrafast bootstrap replicates and 1000 SH-aLRT replicates. The best-fitting model for the final tree was TVM + F + ASC + R2. 2.4. Pan-Genome Analysis The annotated genomes were used for pan-genome analysis to characterize the core (orthogroups present in ≥95% of isolates), accessory (orthogroups present in <95% but ≥1 of isolates) and host-associated genes (FDR/q-value < 0.05 and odds ratios (OR) > 1). A sequence similarity threshold of 95% was used to identify homologous gene clusters among strains. Specifically, for each orthogroup, we compared the gene prevalence within one host group against that across the remaining host cohorts, with the total strain count of each host category set as the background population to calculate corresponding p-values. Fisher’s exact test was used to identify host-associated genes (Table S2). The p-values were corrected using the Benjamini–Hochberg false discovery rate (FDR) method (Table S3). Genes with p < 0.05 and FDR/q-value < 0.05 were considered significantly enriched in the corresponding host. 2.5. Virulence Factor Analysis ABRicate v1.0.1 was used to screen the whole-genome sequences of all strains against the Virulence Factor Database (VFDB; updated April 2025) for virulence factor analysis (Table S4). Alignment hits for virulence factor were filtered using thresholds of ≥60% identity and ≥60% coverage. 2.6. AMR Gene Analysis ABRicate was used to align the whole-genome sequences of all strains against the Comprehensive Antibiotic Resistance Database (CARD; updated April 2025) for AMR gene analysis. Alignment hits for AMR genes were filtered using thresholds of ≥60% identity and ≥60% coverage. ResFinder was used to validate mobile resistance genes. Only resistance determinants using the integrated CARD and ResFinder screening strategy are reported (Table S5). 3. Results and Discussion In this study, 97 duck-derived, 28 chicken-derived, and 33 goose-derived isolates were included. The bacterial pan-genome comprises core and dispensable genomes, representing the conserved and variable components of the gene repertoire, respectively. The analysis identified 5412 gene clusters, including 1469 core-gene clusters, yielding a concatenated core-gene alignment of 1,527,901 bp. Of these, 1423 were present in 99–100% of genomes and 46 in 95–<99%. 3.1. Core-Genome Phylogenetic Analysis The core-genome phylogenetic tree was constructed using single-copy orthologs after recombination filtering. The core-genome phylogeny showed no apparent separation according to host, sampling period, or geographic origin (Figure 1). This finding indicates that there is no clear host-associated phylogenetic separation among the analyzed R. anatipestifer. This is consistent with the findings of previous genomic studies that showed no strict host-associated phylogenetic separation [6]. 3.2. Pan-Genome Analysis Pan-genome analyses revealed several host-associated gene clusters (Figure 2). Cell surface biosynthesis plays a critical role in host adaptation and immune evasion by modulating surface antigenicity, charge, and immune recognition. In chicken-derived strains, the enriched genes included pglH (involved in protein glycosylation), tuaG (glycosyltransferase within the teichuronic acid biosynthetic operon), and wecD/wzxE (polysaccharide biosynthesis-related genes). Genes associated with stress/virulence regulation (ohrR, mazF, symE, parD1/parE1) were also enriched in chicken-derived strains. mazF, parD1/parE1, and symE belong to the toxin-antitoxin (TA) systems, which have been associated with prokaryotic stress tolerance and environmental responses. Notably, the yefM–yoeB system was absent from chicken-derived R. anatipestifer in this study. However, it was enriched in the goose-derived isolates. Nevertheless, owing to the limited number and low genetic diversity of chicken-derived isolates, larger sample sets and targeted PCR assays are needed to further validate the absence of the yefM-yoeB system in chicken-derived strains. Chicken-derived strains displayed significant enrichment of genes involved in DNA repair and recombination (mutS2, mutL) and signal transduction (rcsC) (Figure 2A). The mutS and mutL genes encode core components of the bacterial mismatch repair pathway, which mediates genomic genetic variation and contributes to the maintenance of genomic stability. Metabolism-related genes were enriched in duck-derived strains (Figure 2B). Host-associated genes enriched in duck-derived strains included a complete urease cluster (ureA–ureG) and cofactor assembly factors (hypA and hypB), which are associated with nitrogen metabolism. Functionally, this intact urease system that hydrolyzes urea can promote bacterial survival on host mucosal surfaces, aligning well with the identified host-associated virulence profiles. Meanwhile, duck isolates showed enrichment of zinc uptake genes and the vitamin B12 transporter gene btuF_2, which are associated with metal and nutrient acquisition. Genes related to metabolic and carbon utilization genes were enriched in the goose-derived strains (Figure 2C). These included myo-inositol 2-dehydrogenase encoding gene iolG, and 6-phosphogluconolactonase encoding gene pglA. Thus, DNA repair systems, nutrient acquisition pathway, and TA system-mediated stress regulatory networks showed host-associated distribution patterns. These host-associated genes warrant functional validation to determine their biological significance in R. anatipestifer. 3.3. Virulence Factor Analysis Among all strains, 24 putative virulence genes were annotated using VFDB, with groEL, tufA, and katB being the most common (Figure 3A). The groEL, tufA, and katB were detected in all strains. groEL is involved in the response to environmental stress. tufA contributes to bacterial adhesion, thereby affecting cell invasion and dissemination. katB is associated with protection against peroxide stress. glf and wbtF were present in more than 80% of genomes. These findings indicated that groEL, tufA, and katB are conserved genes in R. anatipestifer. Their conservation across the analyzed isolates warrant further evaluation as potential vaccine-antigen candidates. Several low-frequency putative virulence factors were also identified, including type 8 capsular polysaccharide-related genes cap8E/cap8P/capC, rfbA, pseB, and sdrC and T9SS-related genes sspA/sspB/sspC. Low-frequency virulence factors may be involved in surface-antigen modification and interactions with host immune responses [7]. Variable putative virulence factors (tviB, ureG, and rfbA) were detected in a subset of strains, showing host-associated differences in genomic distribution. Previous studies have shown that alterations in capsule structure in R. anatipestifer significantly impair its survival in the blood, resistance to complement-mediated killing, and pathogenicity [8]. tviB, encoding capsular polysaccharide, was predominantly detected from chicken-derived strains, whereas ureG, part of the urease cluster, was enriched in duck-derived strains and absent in chicken-derived strains. This is consistent with the result of pan-genome host-association analysis. 3.4. AMR Gene Analysis The results showed over 40 putative AMR genes associated with transport systems, multidrug efflux pumps and acquired resistance (Figure 3B). Among the highly prevalent resistance genes, ranA and ranB (predicted ABC efflux pump-encoding genes) were detected in all strains; the β-lactamase-related genes blaOXA-209, blaRAD-1, blaRASA-1 and tet(X) were detected in 155, 82, 76 and 94 strains, respectively. This indicates that R. anatipestifer possesses genomic potential for tetracycline- and tigecycline-resistance. In addition, blaOXA-like and tet(X) have been shown to exhibit high prevalence in R. anatipestifer in previous AMR gene analyses [9]. ErmF, associated with macrolide/lincosamide resistance, was detected in 128 strains, and floR, related to chloramphenicol/florfenicol resistance, was detected in 72 strains. CARD annotation also revealed host-associated differences in the distribution of AMR determinants. The carbapenem-hydrolyzing class D β-lactamase blaRAD-1 and class A extended-spectrum β-lactamase blaRASA-1 were significantly less frequent in chicken-derived strains than in duck- and goose-derived strains [10,11]. Conversely, the dihydrofolate reductase gene dfrA36 was more common in chicken-derived strains than in waterfowl-derived strains (Figure 3B). Differences in antimicrobial use among production systems may have contributed to these patterns, but this possibility cannot be distinguished in the present study from temporal, geographic, or lineage effects. Furthermore, these host-associated AMR determinants require further functional investigation. This study has several notable limitations. Differences in research attention to R. anatipestifer and the regional distribution of waterfowl farming resulted in uneven host sample sizes, geographical heterogeneity, and inconsistent sampling years which may introduce temporal and spatial biases. Moreover, the 118 strain genomes were downloaded from the NCBI database for which antimicrobial exposure, lineage and farm-level confounding factors were unavailable. Accordingly, the present comparative genomic analysis cannot determine the mode of transmission through comparative genomics. Phylogenetic correction for pan-genome-wide association analysis and assessment of phylogenetic robustness will be needed in future studies to reduce confounding phylogenetic effects. Finally, our conclusions are limited to host-associated genomic differences. Further experimental validation is essential to determine the biological functions of these host-associated genes. 4. Conclusions Genomic analysis of 158 R. anatipestifer isolates revealed no clear host-associated phylogenetic separation. Core putative virulence genes were widely conserved, while several putative virulence factors displayed host-associated distribution patterns. The genomes of these isolates harbored multiple putative AMR determinants. Genes involved in cell surface remodeling, TA system-regulated stress response and DNA repair capacity were enriched in chicken-derived isolates, whereas waterfowl-derived isolates showed enrichment of genes associated with nutrient utilization. These host-associated genes represent candidate genomic features for further investigation.