Sustainable antibiotic reduction in poultry production with Pulsatilla saponins and herbal supplementation

Abstract

The prolonged use of antibiotics in poultry production promotes the accumulation and spread of antibiotic resistance genes (ARG), raising concerns for animal health and public safety. Developing effective antibiotic alternatives that support performance while limiting resistance risk is therefore a priority. Using broiler chickens as a model, this study evaluated the effects of Pulsatilla saponins, alone or combined with a compound herbal formulation, on growth performance, immune responses, cecal microbiota, and the intestinal resistome, with an antibiotic-treated group as reference. Growth and immune parameters were integrated with shotgun metagenomic sequencing to characterize microbial and ARG responses to dietary interventions. Compared with antibiotic supplementation, the combination of 0.5% herbal medicine and 0.6% Pulsatilla saponins (ZBZ) combination significantly enhanced immune traits, including spleen index and serum IgA and IgM levels, while increasing cecal microbial diversity and reshaping community composition. Metagenomic analyses showed that antibiotic treatment enriched efflux pump and target modification associated ARG, indicative of a multidrug resistance profile. In contrast, ZBZ markedly reduced the abundance and diversity of multidrug resistance–related ARG. Notably, ZBZ supplementation enriched short-chain fatty acid–producing taxa that were negatively correlated with multiple ARG classes, suggesting that improvements in the intestinal metabolic environment and colonization resistance constrained the expansion of resistant bacteria. Overall, the combined use of Pulsatilla saponins and a compound herbal formulation improved growth and immune performance while reducing intestinal ARG burden through coordinated modulation of the cecal microbiota–resistome axis, providing a sustainable nutritional strategy for antibiotic-reduced poultry production.

Introduction

The poultry industry is one of the fastest-growing sectors of animal agriculture worldwide and plays an indispensable role in ensuring the global supply of animal-derived protein. However, intensive production systems are characterized by high stocking density and elevated stress levels, which substantially increase the risk of infectious diseases in poultry (Maron et al., 2013). For decades, antibiotics have been extensively used as growth promoters and prophylactic agents to improve feed efficiency and reduce mortality rates (Sarmah et al., 2006; Vliex et al., 2024). Nevertheless, prolonged or inappropriate antibiotic use has been shown to disrupt gut microbial homeostasis, accelerate the enrichment and dissemination of antibiotic-resistant bacteria and antibiotic resistance genes (ARG), and threaten animal health and public safety through food-chain and environmental transmission pathways (Jian et al., 2021; Bai et al., 2022). Consequently, the development of safe, effective, and sustainable alternatives to antibiotics has emerged as a central challenge in poultry nutrition and health research.

The gut microbiota functions as a dynamic “microbial organ” that plays a pivotal role in nutrient metabolism, immune regulation, and defense against enteric pathogens (Rooks and Garrett, 2016; Iliev et al., 2025). In chickens, the cecum represents the primary site of microbial fermentation and short-chain fatty acids (SCFAs) production, directly influencing feed energy utilization, muscle deposition, and the maturation and homeostasis of the immune system (Wen et al., 2021). The chicken cecum harbors a complex microbial community dominated by Firmicutes and Bacteroidetes, whose metabolic products are central to fiber fermentation, energy provision, and immune development (Wen et al., 2021). Antibiotic administration disrupts the cecal microbial ecosystem, weakens colonization resistance, and promotes the enrichment and spread of ARG (Yang et al., 2022). These alterations not only impair nutrient utilization and immune homeostasis but also increase the risk of horizontal transfer of resistance genes (Zhang et al., 2021; Yang et al., 2022). Increasing evidence indicates that the gut resistome does not exist in isolation but is tightly coupled with microbial community structure, metabolic function, and host physiological status (Kim et al., 2021; Dongre et al., 2025; Nass et al., 2025). Accordingly, a key unresolved scientific question is how nutritional interventions can be used to modulate the cecal microbiota and resistome in order to improve growth performance while reducing the burden of ARG. However, most existing studies have primarily focused on the effects of antibiotic alternatives on growth performance or immune indices, whereas their systematic impacts on the intestinal resistome and the underlying ecological mechanisms remain insufficiently understood.

Natural plant extracts and traditional herbal formulations are widely regarded as promising alternatives to antibiotics due to their antimicrobial, immunomodulatory, and digestive-promoting properties (Obianwuna et al., 2024; Koorakula et al., 2022; Fonseca et al., 2024). For example, licorice polysaccharides have been shown to enhance immune responses in broiler chickens, increase antibody titers against Newcastle disease, and promote the proliferation of beneficial bacteria such as Lactobacillus and Bifidobacterium (Wu et al., 2022; Lu et al., 2025). Pulsatilla chinensis is rich in triterpenoid saponins and has been reported to possess antimicrobial and anti-inflammatory activities, along with the potential to modulate the gut microbial ecosystem (Zhong et al., 2022). Pulsatilla saponins (PS) have been shown to alleviate disease conditions in mouse models by modulating inflammatory responses and reshaping gut microbial composition, indirectly supporting its role in regulating the intestinal microecology (Xiang et al., 2025). Collectively, these findings suggest that herbal bioactive compounds may improve poultry health and productivity through the remodeling of gut microbial ecosystems. However, most existing studies have focused on the physiological effects of individual plant components, and systematic evidence regarding their influence on microbial functional metabolism and the distribution of ARG in the poultry gut remains limited. In addition, compound herbal formulations, characterized by multi-component and multi-target synergistic actions, may offer advantages over single additives in regulating gut ecological stability; nevertheless, related studies remain scarce, constraining the precise application of herbal extracts in animal nutrition and antibiotic-replacement strategies.

Based on this background, the present study focused on PS and their combination with a compound herbal formulation, establishing a multi-treatment controlled design to systematically evaluate their effects on growth performance, immune function, cecal microbial community structure, functional metabolic profiles, and the intestinal resistome in broiler chickens. Broiler diets were supplemented with different doses of PS and the compound herbal formulation, resulting in nine treatment groups, including a control group, graded PS groups, a herbal formulation group, combination groups, and an antibiotic (kitasamycin) group. At 42 days of age, two optimal “PS + compound herbal formulation” combinations were selected based on growth performance and, together with the control, herbal formulation, and antibiotic groups, constituted five experimental groups for further in-depth analyses (Fig. 1). This work provides microecological and molecular-level evidence supporting the application of plant-derived bioactive compounds in antibiotic-free broiler production and offers a theoretical basis for developing sustainable production systems that balance productivity with ecological safety.

Fig. 1.

A trial on AA broilers fed with the same basal diet supplemented with different feed additives. The treatment groups included: control group (C), 0.3%, 0.6%, 0.9% Pulsatilla saponins (BD/BZ/BG), 0.5% traditional Chinese medicine formula + 0.3%, 0.6%, 0.9% Pulsatilla saponins (ZBD/ZBZ/ZBG), 0.5% traditional Chinese medicine formula (Z), 20 mg/kg kitasamycin (K). A total of 9 groups were included. At 42 days of age, two treatment groups and the C, Z, K group were selected based on the best growth performance, making a total of 5 experimental groups, with 6 chickens (3 males and 3 females) in each group, totaling 30 chickens for subsequent analysis.

Materials and methods

Ethics approval

The animal experimental procedures were carried out in accordance with the Guide for the Care and Use of Laboratory Animals. All experiments involved in this study were approved by the Animal Protection and Utilization Institutional Committee of Yunnan Agricultural University.

Experimental animals and housing management

A total of 162 one-day-old healthy AA broilers with similar physical conditions were randomly assigned to 9 treatment groups, with 3 replicates per group and 6 broilers per replicate, equally divided between males and females. The broilers were raised at the Experimental Poultry Farm of Yunnan Agricultural University for the experiment. Before chick placement, the poultry house and all equipment were thoroughly disinfected by fumigation with formaldehyde and potassium permanganate, and feeding utensils were washed with a 3–5% Lysol solution. Prior to chick arrival, drinking water was pre-warmed to 34–35 °C and supplemented with 8% glucose and multivitamins to alleviate transport stress. Chicks were allowed free access to drinking water immediately after placement and were provided feed 3–5 h later. Birds were reared under ad libitum access to feed and water throughout the experiment. Feeding was conducted three times daily (07:00, 12:00, and 18:00). Diet formulation and nutritional levels were designed according to the Chinese Feeding Standard for Broilers and the NRC (National Research Council) recommendations. Experimental diets were provided in three phases: starter (0–14 d), grower (15–28 d), and finisher (29–42 d) (Table S1). All birds were housed in a multi-tier wire-mesh cage system. Environmental conditions were controlled as follows: brooding temperature was maintained at 35–37 °C during the first week and then gradually decreased by approximately 2 °Cper week until reaching 18–21 °C. Relative humidity was maintained at 65–70% during the first 10 days and approximately 55% thereafter. Ventilation and sanitation were strictly managed. Continuous lighting was provided during the first 48 h, followed by natural daylight supplemented with artificial light during nighttime. Stocking density was adjusted according to bird growth to avoid overcrowding.

Experimental treatments

The basic Chinese herbal formula used in the experiment included Angelica sinensis (19.23%), Lycium barbarum (19.23%), Yam (19.23%), Danshen (3.85%), Schisandra chinensis (19.23%), Rhizoma spp. (6.15%), Osthorium spp. (7.69%), and oyster (5.38%). Previous studies conducted by our research team have shown that this herbal formula can enhance broiler chickens' growth performance without toxic side effects. Pulsatilla saponin (containing 10% triterpene saponins) were obtained from Lanzhou Wotelaisi Biological Technology Co. Ltd. At 42 days of age, based on optimal growth performance, two combined treatment strategies involving PS and the herbal formula were selected. Together with a control group, a traditional herbal medicine group, and an antibiotic group, a total of 5 experimental treatments were established (Fig. 1).

Growth performance measurement

Growth performance was evaluated according to the Technical Regulation for Determination of Broiler Production Performance (NY/T 828—2004). Feed intake was recorded daily by measuring feed offered and feed refusals for each replicate. Birds were weighed weekly to determine body weight (BW) and average daily gain (ADG). Feed conversion ratio (FCR) was calculated as the ratio of feed intake to body weight gain for each replicate over the corresponding period.

Slaughter procedure and sample collection

At 42 days of age, birds were subjected to slaughter and sample collection. Prior to slaughter, birds were fasted for 12 h with free access to water. Birds were humanely euthanized by cervical bloodletting following neck dislocation, a method commonly applied in poultry research and consistent with ethical standards for minimizing animal suffering. Death was confirmed by complete exsanguination.

From each replicate, four birds were randomly selected for carcass trait evaluation (12 birds per treatment, equal sex ratio). Carcass traits, including semi-eviscerated ratio, eviscerated ratio, breast muscle ratio, and thigh meat ratio, were determined according to NY/T 823—2020 (Terminology and Measurement Methods for Poultry Production Performance). Immediately after slaughter, immune organs (spleen, thymus, and bursa of Fabricius) were excised and weighed. Immune organ indices were calculated using the following formulas:

Spleen index (%) = spleen weight (g) / body weight (g) × 100%

Thymus index (%) = thymus weight (g) / body weight (g) × 100%

Bursa index (%) = bursa weight (g) / body weight (g) × 100%

Blood and cecal content sampling

For biochemical and microbiological analyses, two birds were randomly selected from each replicate (six birds per treatment, equal numbers of males and females). Approximately 3–4 mL of blood was collected from the wing vein using sterile syringes and stored at −20 °C until analysis.

After euthanasia, the abdominal cavity was aseptically opened, and the ceca were carefully isolated. Cecal contents (distinct from excreted feces) were gently collected into sterile containers to avoid contamination from intestinal mucosa. Samples were immediately frozen rapidly in liquid nitrogen, transported to the laboratory, and stored at −80 °C until further analysis.

Blood immunomolecular assays

The enzyme-linked immunosorbent assay (ELISA) kits used for the experiments were purchased from Beijing Solarbio Science & Technology Co.,Ltd. The levels of C3, C4, IgA, IgM and IgY in blood were determined by ELISA, and the specific experimental operation was referred to the instructions of the kit.

DNA extraction and sequencing

The SDS method was used to extract total genomic DNA from the samples, excluding host DNA. DNA concentration and purity were assessed using 1% agarose gel, followed by dilution to 1 ng/µL with sterile water. The DNA quantity for each sample was 1 μg for library preparation. Sequencing libraries were prepared using the Illumina DNA library preparation kit. DNA samples were sonicated to 200-500 bp, followed by polymerase chain reaction (PCR). The PCR products were purified, and library size distribution was analyzed using the Agilent2100 bioanalyzer, followed by quantification via real-time fluorescence PCR. Index-encoded samples were clustered on the cBot cluster generation system. The clustered libraries were sequenced on the Illumina platform, generating paired-end reads.

DNA sequence assembly and annotation

The data analysis process, starting from the raw sequencing output, involved optimized procedures such as sequence splitting, quality trimming, and contaminant removal. MEGAHIT was used to perform metagenome assembly (Berendonk et al., 2015), and QUAST software was used to evaluate the assembly results. Coding regions in the genome were identified using MetaGeneMark with default parameters (Wenhan et al., 2010). Redundancy was removed using CD-HIT software, with a similarity threshold of 95% and a coverage threshold of 90% (Fu et al., 2012). Subsequently, optimized sequences were used for assembly and gene prediction, followed by species and functional annotation and classification using databases such as NR, EggNOG, and KEGG. Based on the aforementioned analyses, various statistical analyses and explorations were conducted in multiple directions, including similarity clustering, group sorting, and differential comparison, to validate experimental hypotheses and unveil new insights.

Statistical and network analyses

SPSS 20.0 was used for statistical analysis. Two-tailed unpaired Student’s t-test was used to evaluate the difference between the control group and the experimental. Single-factor analysis and Duncan multiple comparative analysis were used in the growth performance, slaughtering performance and immune indicators. p < 0.05 were considered statistically significantly different. R packages were used to perform pairwise differential analysis between groups to analyze KO, CAZy, and COG. Differentially enriched functional units were screened based on criteria such as logFC and p-value. The R package clusterProfiler was used to enrich the selected functional units, with some of the enrichment results depicted in scatter plots. The R package metagenomeSeq was used to conduct differential abundance gene or functional tests. Normalization methods were used to mitigate the effects of uneven sequencing depth. A zero-inflated log-normal mixture model was also used to address abundance differences in testing caused by inadequate sampling. R software was used to conduct multiple group differential analyses of KO, CAZy, and COG databases via Kruskal–Wallis differential testing. Thereafter, the selected functional units were subjected to enrichment analysis using the clusterProfiler package, and partial enrichment results were visualized in scatter plots.

R software was used to correlate species and phenotype data, examining the correlation between abundance at the genus and species levels and various phenotype data via Spearman analysis to identify classified units with rho > 0.6 and p-value < 0.05. Subsequently, linear fits were plotted. Functional and phenotype data correlation analysis was conducted using R software by examining the association between KEGG pathway, CAZy, and COG abundances and phenotype data via Spearman analysis to identify functional units with rho > 0.6 and p-value < 0.05, followed by plotting linear fits.

Result

The effects of different additives on the growth performance, slaughter performance and immune indicators of AA broilers

With respect to growth performance, Pulsatilla chinensis supplementation markedly improved body weight (BW), feed-to-gain ratio (F/G), and daily weight gain (DWG) in broilers. Compared with the control group, broilers in the BZ, ZBZ, and K groups exhibited significantly higher body weight at 42 days of age, with the ZBZ group showing the most pronounced effect. During both the starter period (1–21 d) and the overall period (1–42 d), the feed-to-gain ratio of the BZ, ZBZ, and K groups was significantly lower than that of the control group. Regarding daily weight gain, the ZBZ group consistently showed significantly higher values than the control group across all growth phases, whereas the BZ and K groups exhibited significant advantages only during specific periods (Table 1).

Table 1.

Effect of PS supplementation on growth performance of broilers.

Items	Group					SEM	)-value
	C	BZ	ZBZ	Z	K
BW	1715.67a	2122.33b	2289.83b	2000.50ab	2070.50b	58.30	0.021
F/G 1-21d	1.56a	1.34b	1.38b	1.54a	1.37b	0.03	0.002
F/G 22-42d	2.12	1.94a	1.90	1.83	1.88	0.04	0.089
F/G 1-42d	1.91a	1.74b	1.72b	1.74b	1.71b	0.02	0.022
DWG 1-21d	27.79bc	30.21ab	33.90a	26.07c	31.62ab	0.87	0.008
DWG 22-42d	50.71a	64.59b	68.19b	64.37b	68.44b	2.14	0.019
DWG 1-42d	39.45a	47.40b	51.05b	45.22ab	50.03b	1.34	0.016

BW, body weight (g); F/G, Feed to gain ratio; DWG, Daily weight gain (g).

Statistical significance was analyzed with Kruskal-Wallis test.

a-cMeans that do not share similar letter in row are significantly different, P < 0.05.

In terms of carcass performance, Pulsatilla chinensis supplementation generally exerted a positive effect on semi-eviscerated ratio (SER) and breast muscle ratio (BMR). The SER of the BZ and ZBZ groups was significantly higher than that of the control group, whereas no significant differences were observed between the Z or K groups and the control. For breast muscle rate, both the BZ and ZBZ groups showed higher values than the control group (Table 2).

Table 2.

Effect of PS supplementation on slaughtering performance of broilers.

Items	Group					SEM	P-value
	C	BZ	ZBZ	Z	K
SER (%)	88.42a	86.43b	87.09b	86.19ab	84.95ac	0.32	0.041
ER (%)	69.44	72.99	73.37	72.73	72.97	0.53	0.118
TMR	22.63	23.11	23.67	21.94	21.92	0.34	0.428
BMR	16.57	18.42	19.73	18.47	18.29	0.36	0.086

SER, Semi-evisceration ratio; ER, evisceration ratio; TMR, Thigh meat ratio; BMR, Breast meat ratio.

a-cMeans that do not share similar letter in row are significantly different, P < 0.05.

With respect to immune parameters, Pulsatilla chinensis supplementation significantly affected the spleen index and selected immunoglobulin levels. The spleen index was significantly higher in the BZ and ZBZ groups compared with the control group. Serum IgA concentration was highest in the ZBZ group and was significantly greater than that in the control and K groups, while IgM levels were also significantly elevated in the ZBZ group. In contrast, no significant differences were detected among groups in bursa index, thymus index, complement components C3 and C4, or IgY levels (Table 3).

Table 3.

Effect of PS supplementation on immune indicators of broilers.

Items	Group					SEM	P-value
	C	BZ	ZBZ	Z	K
Bursal index	0.17	0.16	0.18	0.15	0.16	0.01	0.539
Spleen index	0.10a	0.17b	0.18b	0.11a	0.11a	0.01	≤0.001
Thymus index	0.27	0.36	0.34	0.30	0.28	0.01	0.059
C3	108.85	114.62	117.92	111.97	112.71	1.94	0.692
C4	28.79	30.37	31.75	27.70	28.38	0.62	0.229
IgA	43.41bc	48.97b	53.37a	45.77bc	41.47c	1.12	0.002
IgM	89.47a	87.82a	97.41b	90.02a	90.11a	1.09	0.040
IgY	286.60	304.13	299.21	284.30	289.22	4.77	0.659

a-cMeans that do not share similar letter in row are significantly different, P < 0.05.

The effects of different additives on the microbial diversity in the cecum

Metagenomic sequencing revealed that the cecal microbiota was predominantly composed of bacteria (99.68%), followed by viruses (0.21%), archaea (0.06%), and fungi (0.06%) (Supplementary Fig. 1). Taxonomic annotation against the NCBI NR database identified a total of 153 phyla, 270 classes, 452 orders, 830 families, 2,222 genera, and 8,195 species. At the phylum level, Firmicutes (67.01%) and Bacteroidetes (18.95%) were the most abundant phyla in AA broilers (Fig. 2a), and 135 phyla were shared among the 5 groups (Fig. 2b). Bacterial taxa present in more than 90% of samples were defined as the core gut microbiota, comprising 98 phyla (64.05%), 1,186 genera (53.38%), and 3,904 species (47.64%). Within the core microbiota, 53 phyla, 620 genera, and 1,702 species were detected in all 30 samples, and these 1,702 species accounted for 99.49% of the total relative abundance among the 8,195 identified species (Fig. 2c). Differential taxa among groups were identified using the rank-sum test, followed by linear discriminant analysis (LDA) for dimensionality reduction and assessment of effect size. The results revealed 55 microbial taxa that differed significantly among the 5 groups (Fig. 2d–e). The ZB group exhibited the largest number of differential taxa, including 20 microorganisms such as g_Romboutsia, s_Alistipes sp. 58_9_plus, and s_Lactobacillus agilis, followed by the ZBZ group with 14 taxa, including f__Sutterellaceae, o__Burkholderiales, and c__Betaproteobacteria. The C, K, and Z groups showed 6, 7, and 8 differential taxa, respectively. At the phylum level, 10 phyla differed significantly among the 5 groups. Overall, the C and Z groups clustered closely together, whereas the BZ and ZBZ groups formed a distinct cluster, and the K group clustered separately. In contrast, the K and C groups generally showed lower relative abundances, particularly at the phylum level for Euryarchaeota and Bacteroidota (Fig. 2f). A total of 160 genera differed significantly among groups at the genus level (Supplementary Fig. 2).

Fig. 2.

Comparative analysis of gut microbial composition across chicken groups. a Bacterial community composition at the phylum level. b The Venn diagram illustrates the overlapping relationships among the bacterial phyla. c Core microbial taxa shared between sample groups at three taxonomic resolutions (phylum, genus, species). d LEfSe analysis identified differentially abundant genera as biomarkers based on Kruskal-Wallis test (p < 0.05) with a threshold LDA score >2.0. e Cladogram visualization of taxonomic features (top 50% abundance), with root representing domain Bacteria. Node colors correspond to the genetic line exhibiting maximal abundance at each taxon. Node sizes scale with relative abundance. f Phylum-level differential abundance analysis. Statistical significance was analyzed with Kruskal-Wallis test. Rows that do not share similar letters are significantly different, P < 0.05.

The effects of different additives on the microbial functions in the cecum

KEGG pathway analysis showed that all samples were enriched in six major functional categories, with Metabolism (71.23%), followed by Genetic Information Processing (14.54%), Environmental Information Processing (11.63%), and Cellular Processes (1.37%) (Supplementary Fig. 3a). CAZy analysis revealed that Glycoside Hydrolases (GH), Glycosyl Transferases (GT), Carbohydrate-Binding Modules (CBM), and Carbohydrate Esterases (CE) were abundant across all samples (Supplementary Fig. 3b). At the functional level, COG annotation identified 23 functional categories. The top five categories by average relative abundance were L (Replication, recombination, and repair; 21.13%), G (Carbohydrate transport and metabolism; 12.93%), P (Inorganic ion transport and metabolism; 8.87%), M (Cell wall, membrane, and envelope biogenesis; 6.81%), and J (Translation, ribosomal structure, and biogenesis; 6.77%) (Supplementary Fig. 3c). The differential analysis at KEGG level 3 revealed that the BZ and ZBZ groups were closely clustered together, while the K group treated with antibiotics formed a distinct cluster. Compared with the control group, the BZ and ZBZ groups exhibited higher relative abundances of pathways related to carbohydrate metabolism, fatty acid biosynthesis, and multiple amino acid biosynthesis processes. (Fig. 3a). In contrast, the BZ and ZBZ groups maintained higher abundances of COG categories closely associated with core metabolism, such as glycosidases, aminotransferases, and dehydrogenases with additional enrichment observed in functions related to cell wall biogenesis and cofactor metabolism. Compared with the control group, the BZ and ZBZ groups shared 15 significantly different functional genes, including activities related to formylglycine-generating oxidase, vitamin B12 binding, malate dehydrogenase (NAD⁺), phosphate acetyltransferase, and ABC-type transport systems (Fig. 3b). At the CAZy family level, the K group showed generally lower abundances of multiple GH, GT, and CBM families. The BZ, ZBZ, and Z groups exhibited significant increases in several key GH families involved in the degradation of starch, α-glucans, and cell wall polysaccharides, as well as in selected GT families. Substantial differences in CAZyme profiles were observed among groups, with the C, Z, and K groups clustering together, while the BZ and ZBZ groups formed a separate cluster (Fig. 3c).

Fig. 3.

Comparative analysis of gut microbial functions composition across chicken groups. a KEGG-level3 differential abundance analysis. b COG differential abundance analysis. c CAZy family differential abundance analysis.

Analysis of the association between metagenomics and growth performance and immune function

Gut microbiota profiling demonstrated that different treatments significantly altered the intestinal microbial structure of broilers and were specifically associated with growth and carcass traits, allowing the identification of microbial markers linked to economically important phenotypes. Correlation analysis between body weight and carcass performance indices identified 62 genera at the genus level that were significantly associated with BW, ER, SER, BMR, and TMR, including 10 negatively and 52 positively correlated genera (Fig. 4a). Among these, five genera were identified that showed significant correlations with at least three of the five phenotypic indicators. Mielpintmyces, Parabacteroides, Porphyromonas was significantly negatively correlated with BW, BMR Sanguibacteroides was significantly positively associated with BW, TMR, and ER. Symbiobacterium exhibited significant positive correlations with BW, BMR, and SER (Fig. 4b). Notably, compared with the C and K, in the BZ and ZBZ experimental groups, the negative correlation bacterial genus Mielpintmyces was significantly reduced, while the positive correlation bacterial genera Parabacteroides, Porphyromonas, Sanguibacteroides, and Symbiobacterium were significantly increased (Fig. c). The BZ and ZBZ treatments specifically enriched key microbial taxa associated with muscle development. With respect to BMR and TMR, Sanguibacteroides and Symbiobacterium were identified as core positively correlated genera (Fig. 4b). The ZBZ group exhibited the highest enrichment of these two genera, followed by the BZ group. In sharp contrast, all of the above growth and meat yield associated microbial taxa were markedly suppressed in the K group.

Fig. 4.

Analysis of the association between metagenomics and growth performance. a The Venn diagram illustrates the overlapping relationships among the microorganisms related to growth performance. b Correlation analysis between growth performance and microorganisms. c Analysis of the differences in growth performance among different groups based on microorganisms. Red and gray edges represent positive and negative correlations, respectively, and are scaled by correlation strength (Spearman’s rho > 0.8 or < –0.8; P < 0.05). Nodes represent the 41 ARG type and microbial phylum.

Correlation analysis between key blood immune parameters and gut microbial composition revealed a strong association between the intestinal microbiota and the systemic immune status of broilers. At the genus level, a total of 60 genera were significantly correlated with complement components C3 and C4 and immunoglobulins IgG, IgM, and IgY, including 23 negatively and 37 positively correlated genera (Fig. 5a). Among these, 5 genera were identified as being most strongly associated with immune-related molecules (Fig. 5b). At the genus level, compared with the C and K groups, the negatively correlated genera Desulfonispora and Streptobacillus were significantly reduced in both the BZ and ZBZ groups. Conversely, the positively correlated genera Halarchaeum, Algoriphagus, and Paraburkholderia were significantly enriched in the BZ and ZBZ groups (Fig. 5c). The core associated genera Algoriphagus, Paraburkholderia, and Halarchaeum were identified as the microbial taxa showing the strongest positive correlations with circulating complement proteins C3 and C4 (Fig. 5b). In the BZ and ZBZ groups, both the relative abundance of these genera and the strength of their associations with complement components were significantly higher than those observed in the C and K groups. These findings suggest that the BZ and ZBZ treatments may indirectly enhance complement activation in the innate immune system of broilers by promoting the colonization of specific immune-associated microbial taxa. Further association analysis focusing on humoral immune responses revealed that Desulfonispora and Streptobacillus showed strong correlations with circulating levels of IgY and IgG. The ZBZ group exhibited the most pronounced enrichment of these immune-associated taxa, while the BZ group also showed a clear enrichment trend. In contrast, in the antibiotic-treated K group, the abundance of nearly all beneficial microbial taxa positively associated with C3, C4, IgY, and IgG was markedly reduced. This immunosuppressive microbial shift was consistent with the generally reduced levels of these key immune parameters observed in the serum of broilers from the K group.

Fig. 5.

Analysis of the association between metagenomics and immune function. a The Venn diagram illustrates the overlapping relationships among the microorganisms related to immune indice. b Correlation analysis between immune indice and microorganisms. c Analysis of the differences in immune indice among different groups based on microorganisms. Red and gray edges represent positive and negative correlations, respectively, and are scaled by correlation strength (Spearman’s rho > 0.8 or < –0.8; P < 0.05). Nodes represent the 5 immune indice and significantly changed microbial phylum.

The effects of different additives on the resistance gene groups of microorganisms in the cecum of AA broiler chickens

To systematically evaluate the effects of different treatments on the gut resistome of broilers and its association with the microbial community, we conducted a comprehensive analysis of the functional composition of ARG and their relationships with intestinal microbiota across treatment groups. These ARG represented seven major resistance mechanisms, with antibiotic efflux accounting for the largest proportion (44.25%), followed by antibiotic target alteration (40.19%) and antibiotic inactivation (5.42%) (Fig. 6a). Differential analysis at the resistance mechanism level further revealed functional shifts among treatments, particularly in ARG related to target modification, antibiotic efflux, and antibiotic inactivation. Compared with the C, Z, and K groups, both BZ and ZBZ treatments significantly reduced the relative abundance of ARG associated with antibiotic inactivation and target alteration, with the most pronounced effect observed in the ZBZ group (Fig. 6b). In contrast, the K exhibited a significantly higher relative abundance of ARG related to antibiotic efflux and target alteration compared with the control group (FDR-adjusted p < 0.01). These ARG conferred resistance to 41 classes of antibiotics, with the top 20 most abundant ARG accounting for 90.43% of the total ARG abundance. The predominant ARG included genes conferring resistance to peptide antibiotics (11.66%), fluoroquinolones (9.67%), macrolides (9.83%), tetracyclines (9.34%), penam (5.44%), and glycopeptides (4.62%) (Supplementary Fig. 4). The 5 most abundant resistance categories were cephalosporins, β-lactams, fluoroquinolones, aminoglycosides, peptides, macrolides, and tetracyclines (Fig. 6c). Network analysis revealed significant positive correlations between multiple classes of ARG and several opportunistic pathogens and dominant anaerobes, including Enterococcus, Bacteroides, and Clostridioides (Fig. 6d). The relative abundance of key ARG-associated genera differed significantly among treatments: antibiotic exposure markedly enriched several potential ARG-host taxa, whereas both BZ and ZBZ treatments significantly reduced the abundance of these genera overall (Fig. 6e).

Fig. 6.

Analysis of the Diversity and Variability of ARG. a Composition and abundance of resistance mechanisms within each group. b Differences in antibiotic resistance mechanisms among the groups. c Distribution frequency and average abundance of 1214 ARG across all collected samples. d Analysis of the association network between ARG types and microorganisms. e Microorganisms differential abundance analysis. Red and gray edges represent positive and negative correlations, respectively, and are scaled by correlation strength (Spearman’s rho > 0.6 or < –0.6; P < 0.05). Nodes represent the significantly changed 41 ARG type and 14 significantly changed microbial phylum.

A total of 1,214 ARG were annotated across the 30 samples, of which 1,120 ARG were shared among all samples (Supplementary Figure). Among these, 603 ARG were detected in every sample, accounting for 49.67% of the total ARG repertoire and 98.74% of the overall ARG abundance, indicating that they constituted the dominant resistance gene pool (Fig. 7a). Collectively, the identified ARG conferred resistance to 41 classes of antibiotics, forming 119 resistance combinations, of which 84 ARG–antibiotic combinations (involving 448 ARG) were associated with resistance to two or more antibiotic classes. Principal component analysis (PCA) showed that the first two principal components explained 42.5% of the total variance. The K was clearly separated from the control group along the PC1 axis. In contrast, samples from the BZ and ZBZ groups largely overlapped, with the ZBZ group exhibiting a higher degree of clustering (Fig.7b). Microbiota–ARG co-occurrence network analysis further revealed significant positive correlations between multiple ARG and several opportunistic pathogens or dominant anaerobic taxa, including Enterococcus, Escherichia/Shigella, and Bacteroides. These associations resulted in a markedly increased network connectivity, indicating a more complex resistance linkage pattern (Fig.7c). Heatmap analysis of differentially abundant ARG showed that antibiotic treatment significantly enriched multiple clinically relevant resistance genes. In contrast, the BZ, ZBZ, and Z treatments collectively reduced the relative abundance of these ARG. Among them, the ZBZ group exhibited the lowest levels for several ARG (Fig.7d). Several resistance determinants, including hmrM, MexF, mutant Escherichia coli folP (sulfonamide resistance), and mutant Staphylococcus aureus GlpT (fosfomycin resistance), displayed distinct clustering patterns across treatments. The BZ and ZBZ groups clustered into the same branch, suggesting that different dietary interventions exerted distinct selective effects on resistance gene profiles (Fig. 7d).

Fig. 7.

Analysis of the diversity and variability of ARG. a Principal component analysis of ARG. b number of shared ARG across varying sample sizes. c Analysis of the association network between ARG and microorganisms. d ARG differential abundance analysis. Red and gray edges represent positive and negative correlations, respectively, and are scaled by correlation strength (Spearman’s rho > 0.6 or < –0.6; P < 0.05). Nodes represent the significantly changed ARG and microbial phylum.

Discussion

Herbal medicines and their derived extracts have been widely reported to enhance growth performance in livestock and poultry (Koorakula et al., 2022; Obianwuna et al., 2024). To better elucidate how Pulsatilla-derived extracts and their herbal combinations improve growth performance and immune function in broilers, this study focused on PS alone and in combination with a compound herbal formulation, and systematically evaluated their effects on growth performance, carcass traits, immune responses, cecal microbiota structure, and the intestinal resistome in AA broilers. The results highlighted the pronounced advantages of the ZBZ across multiple production and immune parameters, while also demonstrating its potential to optimize gut microbiota composition and reduce antibiotic resistance gene abundance. The results showed that broilers in the ZBZ group exhibited significantly higher BW, DWG, and improved F/G throughout the entire experimental period compared with the control group, and achieved the highest SER. Shotgun metagenomic analysis revealed that the ZBZ group was enriched in several microbial genera associated with efficient energy utilization and protein deposition, including Parabacteroides, Sanguibacteroides, Porphyromonas, and Symbiobacterium. These microbial taxa have been reported to participate in SCFAs production, particularly acetate and butyrate synthesis, thereby enhancing energy supply to the intestinal epithelium and promoting muscle fiber formation (Xu et al., 2025; Oke et al., 2025). In addition, SCFAs can activate immune-related signaling cascades through G protein–coupled receptors such as GPR41 and GPR43, thereby contributing to immune regulation (Parada Venegas et al., 2019). KEGG functional annotation further indicated that multiple amino acid biosynthesis pathways, including those for lysine, arginine, and methionine, were significantly enriched in the ZBZ group. Amino acids not only serve as direct substrates for muscle protein synthesis but also stimulate myofiber formation through activation of the mTOR signaling pathway (Wu, 2013). Dietary supplementation with these amino acids has been shown to improve feed efficiency and lean meat yield in broilers (Wen et al., 2017). Enrichment of bile acid and fatty acid metabolism may further facilitate the absorption of fat-soluble vitamins, thereby indirectly enhancing production performance. Collectively, these findings suggest that the ZBZ combination may maximize dietary energy utilization, improve nutrient digestibility, and promote muscle deposition through microbiota-driven, multi-layered metabolic networks, achieving growth-promoting effects comparable to those historically obtained through antibiotic supplementation. Nevertheless, further validation using integrated transcriptomic and metabolomic approaches will be required to confirm these mechanistic interpretations.

Gut microbiota interact with host pattern recognition receptors (PRRs), such as Toll-like receptor 4 (TLR4) and NOD-like receptors, through microbial products including metabolites and membrane components, thereby regulating both local and systemic immune homeostasis (Macpherson and Harris, 2004; Belkaid and Hand, 2014). The present study showed that the ZBZ-treated group exhibited significantly increased spleen index as well as elevated serum IgA and IgM levels, and these immune responses were strongly associated with alterations in gut microbial structure, indicating a pronounced microbiota-dependent immunomodulatory effect. Further correlation analyses revealed that the ZBZ and BZ groups were enriched with a set of core microbial genera with potential immune-activating properties, including Paraburkholderia. Certain environmentally derived Paraburkholderia strains have been shown to synthesize immunoactive lipopolysaccharides and antioxidant metabolites, thereby promoting B cell activation and immunoglobulin production while contributing to the modulation of the inflammatory microenvironment in the host (Nešić et al., 2025). Commensal gut microbiota and their structural components, including bacterial membrane lipids, polysaccharides, and lipopolysaccharides, can mediate host immune signaling through PRRs, particularly Toll-like receptors (TLRs), thereby shaping both innate and adaptive immune responses (Di Lorenzo et al., 2019). For example, gut commensals can be recognized via the TLR4/MD-2 complex through their lipopolysaccharides, activating downstream signaling cascades that promote IgA production and B cell differentiation in gut-associated lymphoid tissues, while also regulating inflammatory pathways such as NF-κB and maintaining immune tolerance. This microbiota–immune interaction framework has been summarized in multiple reviews, highlighting the complex bidirectional communication between the gut microbiota and host immunity and its role in maintaining immune homeostasis (Valentini et al., 2014; Yoo et al., 2020). Correlation analyses further demonstrated that the abundance of these dominant genera was positively associated with complement components C3 and C4 as well as IgA and IgM levels, suggesting that plant-derived bioactive compounds may indirectly enhance host immune defenses by reshaping the gut microbial ecosystem and promoting the colonization of immunomodulatory functional bacteria. These findings are consistent with previous reports indicating that plant-based feed additives exert their immunomodulatory effects through a “microbiota–immune axis” (Nantapo et al., 2025; Sahoo et al., 2025). Improvements in nutrient metabolism mediated by the gut microbiota, particularly involving branched-chain amino acids, fatty acids, and vitamins, can provide immune cells with sufficient biosynthetic substrates and energy supply, and such metabolic–immune coupling has been recognized as a central mechanism underlying the growth-promoting effects of plant-based additives (Yahsi and Gunaydin, 2022). Several essential amino acids act as key nutrient-sensing signals for the mTOR signaling pathway and are capable of regulating immune cell activation, proliferation, and differentiation. Previous studies have demonstrated that amino acids influence T cell fate decisions through Rag GTPase-mediated activation of mTORC1, promoting differentiation toward effector phenotypes and supporting the metabolic demands of immune responses (Ren et al., 2017; Yang et al., 2023). Vitamin B12 and folate are essential for DNA synthesis and ensure proper immune cell proliferation, while enhanced antioxidant capacity can mitigate immune cell damage under inflammatory conditions (Gombart et al., 2020; Behringer et al., 2025). This mode of action, whereby gut microbiota facilitate the metabolism of key nutrients, suggests that the efficacy of ZBZ relies not only on direct microbial immunomodulation but also on sustained immune enhancement through metabolic–immune interactions. In addition, certain bacterial taxa enriched in the ZBZ group are capable of bile acid metabolism, leading to the activation of nuclear receptors such as the farnesoid X receptor (FXR), which indirectly contributes to inflammation suppression and maintenance of immune homeostasis (Zeng et al., 2025).

In intensive poultry production, the long-term or high-intensity use of antibiotics continuously selects the intestinal microbiota, promoting the rapid enrichment of antibiotic-resistant bacteria and ARG, and facilitating their spread between the microbial communities and ecological niches through horizontal gene transfer (HGT). This is the core mechanism of the expansion of intestinal antibiotic resistance genomes and an important source of the spread of antibiotic resistance (Thomas and Nielsen, 2005; Williams et al., 2023). The results of the metagenomic analysis in this study showed that the group K exhibited a marked reduction in gut microbial diversity, accompanied by pronounced functional shifts, with the ARG profile dominated by efflux pump–related genes, including macB, mepA, acrB, and tolC. Extensive evidence indicates that efflux pump systems can confer resistance to multiple classes of antibiotics simultaneously and constitute a key mechanistic basis for the development and persistence of multidrug resistance (MDR), particularly under prolonged antibiotic exposure where they are strongly favored by positive selection (El Ghallab et al., 2024; Shen et al., 2025; Piddock, 2006). In addition, the abundance of ARG associated with antibiotic target modification, such as vanXYC, fabG, and rpsA, was concurrently elevated in the K group, suggesting that under sustained drug pressure, gut microbial communities rely not only on active efflux but also increasingly adopt strategies involving alterations in cell wall synthesis, fatty acid metabolism, or ribosomal structure to evade antibiotic inhibition. This combination of resistance mechanisms is highly consistent with findings from multiple poultry and swine farm surveys, which demonstrate that prolonged antibiotic use drives resistance profiles to shift from single-drug resistance toward broad-spectrum, multi-mechanistic resistance dominated by efflux pump–mediated MDR (Enshaie et al., 2025; Zalewska et al., 2021). The relative abundances of multiple ARG categories were significantly reduced in the ZBZ and BZ treatment groups, particularly multidrug resistance regulatory genes such as evgA, nalC, and lsaB, as well as several genes involved in antibiotic inactivation, including aadA27, SAT-4, and SPG-1. This reduction is unlikely to result from a simple antimicrobial effect but rather reflects a broader restructuring of the gut microbial ecosystem. The inclusion of plant-derived bioactive compounds significantly increased microbial diversity and functional redundancy, allowing ARG-free functional taxa to gain a competitive advantage in ecological niche occupation, thereby weakening the relative dominance of resistant strains, a process consistent with the theoretical framework of the “ecological dilution effect” (Forslund et al., 2013). Previous studies have shown that SCFAs, as major metabolites generated through microbial fermentation, enhance colonization resistance and suppress the expansion of pathogenic bacteria in the gut. The SCFAs, including acetate, propionate, and butyrate, significantly inhibit the growth of pathogens such as Enterobacteriaceae under physiological pH conditions and help maintain the colonization advantage of beneficial microbial communities (Chang et al., 2024). In addition, SCFAs improve intestinal barrier function by enhancing tight junction protein expression and epithelial integrity, thereby reducing opportunities for pathogen invasion (Pérez-Reytor et al., 2021; Ma et al., 2021). Therefore, ZBZ treatment may indirectly constrain the spread of ARG-hosting bacteria through coordinated regulation of the metabolic environment and restructuring of microbial competition. Moreover, multiple bioactive compounds present in PS and compound herbal formulations, including saponins, polyphenols, and alkaloids, have been reported to exert selective antimicrobial effects by suppressing the overgrowth of key potential ARG hosts such as Enterococcus faecium and Bacteroides fragilis, thereby maintaining microbial stability under low resistance pressure (Shanmugam et al., 2025). In the present study, ZBZ and BZ treatments showed a pronounced increase in the functional potential of polysaccharide-degrading enzyme families (GH and GT), accompanied by a concurrent decrease in ARG abundance, further supporting the ecological association between high functional diversity and a reduced resistance burden. Compared with previous studies that primarily focused on single plant extracts, the present findings demonstrate that the combination of PS with compound herbal formulations is more effective in reducing both the diversity and abundance of ARG. This advantage likely arises from synergistic interactions among multiple components. Firstly, different bioactive compounds exert additive or complementary inhibitory effects on resistant bacteria. Additionally, the combined formulation more effectively promotes the establishment of a structurally stable, functionally diverse, and low-risk intestinal ecosystem at the community level. This multi-target and multi-level regulatory mode has greater application potential in controlling drug resistance compared to a single additive.

In conclusion, the combination of 0.5% herbal medicine and 0.6% PS not only has the potential to replace antibiotics at the production performance level, but also can significantly reduce the abundance of bacterial populations carrying ARG as well as the types and quantities of ARG, thereby reducing the enrichment and transmission risks of drug-resistant genes in the intestinal tract. Previous studies have also shown that compared with chlortetracycline, plant-derived benzylisoquinoline alkaloids can significantly reduce the relative abundance of ARG in the intestinal tract of poultry (Spiljar et al., 2017). Therefore, the ZBZ combination provides a feasible nutritional intervention path for building a low-resistance burden poultry breeding system in the "post-antibiotic era". Although this study clearly identified the multi-dimensional advantages of the ZBZ combination, there are still some shortcomings. This study did not further validate the functional prediction results of the intestinal microbiota by combining metabolomic or transcriptomic data. Additionally, the reproducibility verification under large-scale rearing conditions was insufficient. Therefore, future research can jointly analyze the mechanism of action of the PS-complex combination using multi-omics approaches, and conduct multi-site validations across different breeds and rearing systems. This will help to build an economic benefit assessment model and provide data support for industrial promotion.

Conclusion

The combination of PS and traditional Chinese herbs (ZBZ) has demonstrated unique and significant effects in enhancing the production performance of broilers, improving immune function, optimizing the structure of the intestinal microbiota, and reducing antibiotic-resistant genes. This strategy not only can replace antibiotics in short-term performance but also has long-term significance in terms of health safety and sustainable farming. Priority can be given to promoting antibiotic-free farming.

Author contributions

Z.J., T.D. conceived and designed the project. Z.J. analyzed the data and the images in the manuscript. Z.J., R.Z., S.H., X.H. collected the metagenomic data. R.Z., Y.Y., K.W., C.G., J.J. and Y.H. provided invaluable feedback and insight for the analysis. Z.J. and T.D. wrote the initial manuscript. All authors approved the final version of the manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Zonghui Jian: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Ruohan Zhao: Methodology, Investigation, Formal analysis. Xiannian Zi: Supervision, Software, Formal analysis, Data curation. Shichun He: Writing – original draft, Software, Formal analysis, Data curation. Xiaoming He: Visualization, Validation, Formal analysis. Yanlin Ye: Writing – original draft, Visualization, Validation. Kun Wang: Writing – original draft, Methodology, Investigation. Changrong Ge: Writing – review & editing, Visualization, Validation. Junjing Jia: Writing – review & editing, Visualization, Validation. Yuanyuan Hu: Supervision, Software, Data curation. Tengfei Dou: Writing – review & editing, Writing – original draft, Methodology, Investigation, Data curation, Conceptualization.

Disclosures

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.