2025 Volume 62 Article ID: 2025028
Growth performance, an important trait in the broiler industry, is defined by both the host genome and gut microbiota. At present, it is not known how gut microbiota contribute to the growth of Dahen broilers, a commercially important breed in China. In this study, we used metagenome sequencing to compare the taxonomic composition and functional implications of cecal microbiota in fast-growing Dahen broilers and slow-growing Tibetan chickens. A total of 2,207,811 unique genes were assembled in the non-redundant set, and 99% of them were taxonomically annotated as having a bacterial origin. The fast-growing group displayed a higher alpha diversity than the slow-growing group in terms of ACE, Chao1, and Good’s coverage statistics. The two groups presented also significantly different (P < 0.05) relative abundances of the genera Collinsella, Olsenella, Pyramidobacter, Basidiobolus, and Mieseafarmvirus, along with that of eight species (e.g., Olsenella timonensis and Victivallis sp. Marseille Q1083). Although not statistically significant, we found a higher expression of several energy metabolism-related eggNOG terms in the fast-growing group. In summary, the present study identifies gut microbiota associated with growth performance in Dahen broilers and offers new tools for gut microbiome-related intervention in this breed.
Broiler growth rates increased by more than 4-fold between 1957 and 2005 due to genetic and non-genetic improvements[1,2]. This economically important trait is characterized by low to moderate heritability, depending on age and breed[3,4]. Growth performance is closely related to feed efficiency in broilers[5] and other livestock species[6]. Using genomic information, Li et al. (2021) reported moderate to strong genetic correlations between growth rate and feed efficiency in purebred broilers[7]. Both traits are of economic importance and contribute to carbon emissions[8].
Besides the host genetic background, gut microbiota represent another important driver of individual growth[5,7,9]. Oakley et al. (2014) summarized the spatiotemporal variability of gut microbiota in chickens and the consequent effect on host health and nutrition[10]. Overall, gut microbiota provide diverse benefits for the host, such as reducing pathogenic bacteria via protective barriers and/or competitive colonization, regulating immune responses, facilitating dietary digestion, and producing additional nutrients[11]. The effects of dietary supplementation with zinc bacitracin and avilamycin on broiler growth are mediated by altered composition and diversity of gut microflora[12]. The latter are responsible also for the negative effect of high environmental ammonia content on broiler growth performance[13]. Additionally, different rearing systems can significantly influence the growth, gastrointestinal development, and gut microbiota of broilers[14]. The above evidence hints at a strong interdependence between gut microbiota and growth performance in broilers.
Owing to the widespread application of high-throughput sequencing, it is now easier to investigate microbial composition and function under diverse conditions[15]. First, partial or full sequences of the 16S rRNA gene can be obtained in high-throughput and cost-effective ways, after which they can be used to detect different bacterial species or phylotypes[16]. Second, whole genomes can be collectively sequenced (i.e., shotgun metagenome), and microbial genes can be assembled, taxonomically annotated, and functionally predicted[17]. Peterson et al. (2021) assessed the advantages of 16S rRNA and metagenome sequencing to profile gut microbiomes[18]. In general, shotgun metagenome sequencing yields functional gene profiles and achieves a much higher resolution of taxonomic annotation[19]. Therefore, it has become a powerful tool for exploring microbial communities under various conditions, especially because of significantly lower sequencing costs and rapid advances in bioinformatics pipelines[20]. Dahen broilers have been genetically selected for their specific meat flavor; however, their growth and meat production can be further improved[21]. In this study, we used metagenome sequencing to explore gut microbiota associated with differential growth performance in fast-growing Dahen broilers as opposed to slow-growing Tibetan chickens. Results reveal the contribution of gut microbiota to host growth.
This study was approved by the Animal Care and Use Committee of the Sichuan Animal Science Academy (202214865). All efforts were made to minimize the suffering of experimental animals.
Animals and sample collectionEight Dahen S07 broilers and eight Tibetan chickens from Sichuan Dahen Poultry Breeding Co., Ltd. were fed under the same conditions with free access to food and water. Prior to the age of four weeks, all broilers were housed in floor pens with fresh wood sawdust. The stocking density was approximately 15 birds per m2, with temperature of 16–27°C and a 16/8 h light/dark photoperiod. Between the ages of four and seven weeks, broilers were housed individually and fed ad libitum with water and a commercial pellet diet. The metabolizable energy at 1–21 days of age was 12.14 MJ/kg, at 22–42 days of age it was 12.45 MJ/kg, and at 43–49 days of age it was 12.91 MJ/kg. At seven weeks of age, the average body weights (± standard deviation) of fast-growing and slow-growing groups were 1251.9 ± 7.0 g and 494.4 ± 9.0 g, respectively (P = 1.15E-15 with Student’s t-test). The chickens were healthy and had no history of infectious disease. Animals were sacrificed via exsanguination under deep anesthesia following an intravenous injection of sodium pentobarbitone (>35 mg/kg body weight). Cecum samples were collected from each individual, immediately frozen in liquid nitrogen, and stored at -80°C until DNA extraction. To reduce the risk of Type II errors, Select Statistical Services (https://select-statistics.co.uk/calculators/sample-size-calculator-two-means/) was used for power analysis of the sample.
Metagenomic sequencing and quality controlsMicrobial genomic DNA was extracted from 200 mg of frozen cecal contents using the QIAamp® Fast DNA Stool Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s protocol. The concentration, quality, and integrity of the genomic DNA were evaluated using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis. Paired-end libraries with an insert size of 350 bp were constructed using the following steps: genomic DNA fragmentation, end repair, A-tailing, DNA adapter ligation, and PCR amplification (Novogene, Tianjin, People’s Republic of China). DNA libraries were finally subjected to 150-bp paired-end sequencing on a NovaSeq 6000 platform (Illumina, San Diego, CA, USA).
Using the fastp tool[22], raw sequencing reads were discarded if they contained more than 10 ambiguous bases (N), more than 40 low-quality bases (Q value <38), or an adapter sequence (>15-bp overlap). Among quality-controlled reads, potential contamination of host DNA was further removed by mapping all reads against chicken reference genome sequences (bGalGal1.mat.broiler.GRCg7b) using the Bowtie2 tool and default parameters[23]. During this process, BBmap 38.93-0 was used to align the reads against the host reference genome. Successfully aligned reads were subsequently filtered out using default parameters for processing. The software operating parameters were as follows: nodisk k=13 minid=.90 usemodulo=t fast=t noheader=t notags=t.
Gene category and annotationFor each sample, clean sequencing reads were first assembled into a scaffold sequence using the MEGAHIT tool and the “-presets meta-large” parameter[24], upon which continuous sequences within the scaffolds (i.e., contigs) were obtained. After discarding short contig sequences of <500 bp in length, genes were predicted using the Prodigal tool and the “-p meta” parameter[25]. Predicted open reading frames (ORFs) shorter than 100 bp were removed, and a non-redundant gene catalog of all samples was constructed using the CD-HIT tool and “-c 0.95, -G 0, -aS 0.9, -g 1, -d 0” parameters[26]. Clean sequencing reads were remapped to these non-redundant gene sequences using the Bowtie2 tool and default parameters[23], allowing for quantification of gene expression in each sample[27].
All non-redundant genes were taxonomically annotated by searching against the NCBI non-redundant protein database using the DIAMOND tool and “blastp, evalue < 1e-5” parameter[28]. The lowest common ancestor algorithm was employed to resolve any non-consensus annotation. Similarly, non-redundant genes were functionally annotated by searching their evolutionary genealogy in Non-supervised Orthologous Groups (eggNOG), Kyoto Encyclopedia of Genes and Genomes (KEGG), Carbohydrate-Active Enzymes Database (CAZy), and Comprehensive Antibiotic Resistance Database (CARD)[29,30,31,32].
Taxonomic composition and functional implicationsBased on unique assembled genes, three alpha diversity indices (ACE, Chao1, and Good’s coverage) were calculated for each sample using the VEGAN R package[33]. Principal component analysis (PCA) was performed using STAMP software to cluster the samples based on their abundance at class level[34]. Differentially abundant genera and species between the fast-growing and slow-growing groups were tested using Welch’s t-test (P < 0.05). Based on gene expression profiles at the phylum, genus, and species levels, intergroup differences were tested by permutational multivariate analysis of variance using the VEGAN R package[33]. After obtaining the annotated functional terms from eggNOG, KEGG, CAZy, and CARD, differential abundances between the fast-growing and slow-growing groups were tested using the Wilcoxon rank-sum method and were considered significant if the false discovery ratio was <0.05.
Statistical AnalysisExcel 2019 (Microsoft, Albuquerque, New Mexico, USA) was used for statistical analysis and significance testing of the data.
On average, 108 million clean reads were obtained per sample after quality control, and 3.88% were discarded because of host DNA contamination (Table S1). A total of 3,495,974 contigs were successfully assembled, with average total size of 269 Mbp, N50 length of 1465 bp, and N90 length of 586 bp (Table S2). The assembled contigs harbored 5,649,777 predicted ORFs, with average length of 661 bp and maximum length of 23,889 bp. After removing redundant sequences, 2,207,811 unique genes were obtained in the non-redundant set, with average GC content of 45.4% and sequence length of 679 bp (Fig. 1A). A total of 3037 unique genes were discarded because either poorly expressed (<3 reads) or detected in only one sample. The gene expression pattern was comparable across all samples, except for sample “A2” in the fast-growing group (Fig. 1B). The 16 samples contained 202,725 core and 2,204,774 pan genes (Fig. 1C). Given that sample “A2” failed to meet the inclusion criteria, it was excluded from subsequent analyses.
Distribution of sequence length (A) and expression level (B) for the assembled unique genes, and the observed number of core and pan genes (C). Fast-growing group: A1–A8 and 1–8; slow-growing group: B11–B18 and 9–16.
Up to 99% of unique genes were taxonomically annotated as having a bacterial origin. The fast-growing group exhibited greater alpha diversity than the slow-growing group in terms of ACE, Chao1, and Good’s coverage (Fig. 2A). However, intra-group variability was also high, which was confirmed by poor separation of fast-growing and slow-growing individuals in PCA clustering of whole-gene expression at the class level (Fig. 2B). Linear discriminant analysis (LDA) effect size (LEfSe) analysis was conducted to ascertain any substantial disparities in the relative abundances of microbial taxa between chickens belonging to different growth groups, with the aim of utilizing these as biomarkers. A total of 62 biomarkers with linear discriminant analysis scores >2 were identified (Table S3). Six biomarkers (Lactobacillus agilis, Lachnoclostridium sp An14, Clostridium sp CAG 306, Bacteria, Phascolarctobacterium faecium, and Alistipes dispar) were identified in the slow-growing group and 56 (e.g., Prevotella sp AG 487 50 53 and Sanguibacteroides justesenii) were identified in the fast-growing group (Fig. 3). No significant intergroup differences were observed upon variance analysis of expression for all taxonomically annotated genes (Table 1). Instead, five genera (Collinsella, Olsenella, Pyramidobacter, Basidiobolus, and Mieseafarmvirus) and eight species (e.g., Olsenella timonensis and Victivallis sp. Marseille Q1083) were significantly differentially abundant between the fast-growing and slow-growing groups (P < 0.05; Fig. 4).
Alpha diversity revealed by unique genes (A), NMDS-based clustering of samples by abundance at class level (B), PCA-based clustering of samples by abundance at class level (C), and principal coordinate analysis-based clustering of samples by abundance at class level (D).
LEfSe analysis of cecal microbiota.
| Levels | Variance analyses | |||||
| Items | Df | SumOfSqs | R2 | F statistic | P value | |
| Phylum | Group | 1 | 0.2161 | 0.09574 | 1.3765 | 0.120 |
| Residual | 13 | 2.0410 | 0.90426 | — | — | |
| Total | 14 | 2.2571 | — | — | — | |
| Genus | Group | 1 | 0.18851 | 0.08801 | 1.2545 | 0.233 |
| Residual | 13 | 1.95349 | 0.91199 | — | — | |
| Total | 14 | 2.14200 | — | — | — | |
| Species | Group | 1 | 0.21601 | 0.09568 | 1.3755 | 0.112 |
| Residual | 13 | 2.04150 | 0.90432 | — | — | |
| Total | 14 | 2.25751 | — | — | — | |
Note: Df, degrees of freedom; SumOfSqs, sum of squares of deviations; R2, proportion of explained variance.
Significant differential abundance of five genera (A) and eight species (B) between fast-growing and slow-growing groups.
Of all the unique genes, 70.5%, 41.5%, and 2.5% were functionally annotated in the eggNOG, KEGG, and CAZy databases, respectively. Similar to taxonomic composition, individuals in the fast-growing and slow-growing groups were not well separated according to overall abundance patterns of all annotated functional terms (Figs. S1, S2, and S3). Although no significant difference existed (after multiple comparisons and adjustments), a trend towards greater abundances in the fast-growing group was identified for several energy metabolism-related eggNOG terms (P > 0.05, Fig. S4), such as ATP-binding protein (COG0489), displays ATPase and GTPase activities (COG1660), gluconate (COG2610), 2-hydroxyglutaryl-CoA dehydratase (COG1775), GntR family transcriptional regulator (COG2186), peptidase m42 family protein (COG1363), and permease (COG2233). Similarly, four KEGG pathways (endocytosis, GnRH signaling pathway, fluorobenzoate degradation, and the Ras signaling pathway) showed differential abundance between the fast-growing and slow-growing groups. Unigene sequences were aligned to the protein sequences in the CAZy database and classified into six enzyme classes. The two most abundant classes in all samples were glycoside hydrolases and glycosyltransferases (Fig. 5). However, no differences were observed between the two groups with respect to functionally annotated CAZy terms (Fig. S5). Finally, sequencing data were annotated for CARD-resistance gene functions. The three antibiotic resistance ontology names exhibiting the highest proportion of genes were rpoB2 (791 genes), lima 23S ribosomal (237 genes), and Bifidobacterium adolescentis rpoB (160 genes) (Fig. 6A). Four biomarkers were identified by LEfSe analysis, and the LDA score was >2 (Table S4). Diaminopyrimidine antibiotic, glycylcycline, triclosan, and nitroimidazole antibiotic were characteristic of the fast-growing group (Fig. 6B).
Functional composition of genes in the CAZy database for fast-growing and slow-growing groups.
Visualization of functional annotation results of CARD resistance genes. Top 10 antibiotic resistance ontology display images annotated through the CARD database (A), and LEfSe analysis of cecal microbiota (B).
In livestock, gut microbiota are involved primarily in dietary digestion and absorption; however, their metabolites can regulate also host immune responses and other physiological conditions[35]. Prebiotic supplementation has become increasingly popular in animal husbandry to promote dietary intake, growth, and health via manipulation of gut microbiota composition and function[36]. Advances in high-throughput sequencing and bioinformatics have made it possible to comprehensively explore the association between gut microbiome and growth performance in poultry[37,38], pigs[39,40], and cattle[41]. Using metagenome sequencing, we compared gut microbiota composition and function in relation to growth rate of Dahen broilers, a commercial breed in China.
The profiling of intestinal digesta samples in chickens allowed Huang et al. (2018) to establish a comprehensive gene catalog of gut microbiota, as well as identify distinctive characteristics and temporal changes in foregut and hindgut microbiota[37]. Yang et al. (2022) recently investigated dynamic changes in gut microbiota and function during growth of broilers and found a gradual decrease in alpha diversity; although beta diversity differed significantly across time points[42]. Similarly, 16S rRNA gene sequencing in chickens revealed a clear difference in cecal microbial community between the duodenum and feces[43]. Accordingly, we selectively collected cecal samples and subjected them to metagenomic sequencing. Based on integrated analyses of both gut microbiota and gut-host metabolites[44], the relative abundances of gut microorganisms differed significantly between the two broiler breeds, which presented distinct fat deposition and growth rates. Differences in gut microbiota composition and diversity have been observed between high- and low-weight broilers[42]. In laying hens, cecal microbiota composition was found to significantly affect feed efficiency[43]. Contrary to previous reports, the present study detected no significant disparity in alpha diversity among fast- or slow-growing groups. However, individuals with higher growth rates tended to demonstrate greater diversity. Therefore, the differential composition of gut microbiota may be related to factors other than growth rate in chickens.
An abundance of Lactobacillus and Akkermansia in the gut microbiota has been suggested to improve feed efficiency in laying hens[43]. Moreover, 22 genera (e.g., Romboutsia, Corynebacterium 1, and Gallibacterium) are likely associated with growth performance through their influence on immune responses and energy metabolism in broilers[42]. In Xiayan chickens, an indigenous breed in China, the relative abundance of Lactobacillus is positively associated with feed efficiency[45]. Dietary supplementation of plant-derived growth promoters favors this genus, thereby improving growth performance in broilers[37]. In this study, we found that the relative abundances of five genera and eight species differed significantly between fast-growing and slow-growing groups; specifically, the genus Olsenella was associated with significantly better growth. Olsenella has been consistently linked to feed efficiency in sheep[46] and contributes to growth under different feeding regimes in rabbits[47]. Conversely, a reduced abundance of the genus Collinsella was reported in pigs with lower residual feed intake[48], which is in contrast to the observations from this study. Hence, feed efficiency may depend on several factors, including variety, environment, and nutrition.
In addition to taxonomic composition analyses, metagenome sequencing data can be used to comprehensively assemble genes and vastly facilitate the functional investigation of microbiota[49,50]. Based on functional mapping of assembled genes, Du et al. (2020) found that KEGG terms associated with genetic information processing were more abundant in individual Xiayan chickens with higher feed efficiency; whereas genes involved in amino acid, fatty acid, amino sugar, and nucleotide sugar metabolism were less frequent[45]. Functional prediction using 16S rRNA gene sequencing in chickens identified protein and amino acid metabolism as the most dissimilar with respect to feed efficiency[43]. Yang et al. (2022) found that metabolism, genetic information processing, and environmental information processing might affect growth performance of broilers by regulating carbohydrate and lipid metabolism[42]. The present study found a significant association between energy metabolism or signaling processing-related functions and growth performance, which is consistent with previous studies[42,43]. Here, the two most abundant enzyme types were glycoside hydrolases and glycosyltransferases. However, no differences between the two groups were observed for CAZy functional annotations.
To further characterize the functional differences between microbiomes associated with divergent growth phenotypes, we profiled antibiotic resistance genes, which might reflect ecological adaptations or indirect effects on host physiology. The three antibiotic resistance ontology names exhibiting the highest proportion of genes were rpoB2 (791 genes), lima 23S ribosomal (237 genes), and Bifidobacterium adolescentis rpoB (160 genes). Diaminopyrimidine, glycylcycline, triclosan, and nitroimidazole antibiotics were characteristic of the fast-growing group. Consequently, although not primary growth-related features, these elements provide a more comprehensive characterization of functional differences between microbiomes associated with divergent growth phenotypes. These results offer a foundation for future mechanistic studies exploring how specific microorganisms influence host physiology beyond direct nutrient provision.
It is important to note that this study utilized Dahen and Tibetan chickens as model systems to uncover functional microbiome features linked to divergent growth phenotypes within a controlled genetic framework. Although the core microbial functions identified (e.g., enhanced nutrient metabolism in fast growers) represent potential mechanisms relevant for growth, the effect of genetics on the specific microbial taxa or gene variants cannot be excluded. Consequently, the direct extrapolation of these specific microbial signatures to commercial poultry lines requires further validation in target breeds under relevant production conditions. It is recommended that future studies examine whether the identified functional pathways are indeed conserved drivers of growth differences across diverse chicken genotypes. Research on gut microbiota of Dahan broilers is currently in its infancy. A comparison of the cecal microbiota in Tibetan and Qingyuan chickens revealed discrepancies in microbial composition and abundance[51], in line with the results of the present study and the existence of breed-specific characteristics. Future research on Chinese chicken breeds will provide further insight on the composition and function of gut microbiota.
In this study, we applied metagenome sequencing to investigate the taxonomic composition and functional implications of gut microbiota associated with differential growth performance in Dahen broilers and Tibetan chickens. We found a clear association between growth rates and the abundance of specific gut microorganisms. Two chicken breeds presented significantly different relative abundances of the genera Collinsella, Olsenella, Pyramidobacter, Basidiobolus, and Mieseafarmvirus. The hypothesis that microbial diversity, rich subtypes, and biological functions have a direct impact on the growth performance of broilers has yet to be substantiated. Future research may reveal the molecular mechanisms underlying the growth of these breeds and facilitate the development of more efficient broiler breeding programs.
This study was supported by the Sichuan Province Innovation Team Project (SCCXTD-2025-24) and National Modern Agricultural Industrial Technology System Construction Project (Grant No. CARS-41), Sichuan Science and Technology Program (2021YFYZ0031), and Sichuan Provincial financial operation special project (SASA2025CZYX002).
Jiangxian Wang: Conceptualization, Data curation, Formal analysis, Writing-original draft, Writing-review and editing. Chunlin Yu: Conceptualization, Data curation, Formal analysis, Writing-original draft, Writing-review and editing. Mohan Qiu: Conceptualization, Data curation, Formal analysis, Investigation, Project administration, Writing-original draft. Xia Xiong: Data curation, Formal analysis, Investigation, Project administration. Han Peng: Formal analysis, Investigation, Methodology. Shiliang Zhu: Formal analysis, Investigation, Methodology, Supervision. Jialei Chen: Investigation, Methodology, Project administration, Software. Xiaoyan Song: Investigation, Methodology, Project administration, Resources. Chenming Hu: Investigation, Methodology, Project administration, Resources. Bo Xia: Methodology, Project administration, Resources, Software, Supervision. Zhuxiang Xiong: Software, Supervision, Validation, Visualization. Longhuan Du: Software, Supervision, Validation, Visualization. Chaowu Yang: Project administration, Validation, Visualization, Writing-review and editing, Funding acquisition. Zengrong Zhang: Resources, Validation, Visualization, Writing-review and editing, Funding acquisition. All authors interpreted the results, edited the manuscript, and approved the final manuscript.
The authors declare no conflict of interest.
