Microbiome divergence by rearing system: Grazing sheep harbor fiber-degrading Bacteroidetes, whereas confinement favors opportunistic Acinetobacter and enrichment of clinical ARGs (MexK/adeJ).

Antimicrobial resistance (AMR) has emerged as one of the most pressing global health threats, undermining the efficacy of antibacterial therapies as pathogens acquire and disseminate resistance determinants (Kariuki, 2024). Livestock production systems contribute substantially to this crisis, accounting for approximately 70% of global antimicrobial consumption (Van Boeckel et al., 2017). The animal gut microbiome represents a particularly important reservoir of antibiotic resistance genes (ARGs), where dense and diverse microbial communities promote horizontal gene transfer (HGT) via mobile genetic elements (MGEs), enabling the spread of resistance from commensals to pathogens (Hu et al., 2013). Mounting evidence indicates that livestock-associated ARGs can cross ecological boundaries, reaching humans through direct contact, food production chains, and environmental pathways (Larsson and Flach, 2022).

The sheep gut harbors an exceptionally dense and diverse microbiota, creating a hotspot for ARG acquisition and transfer (Kamke et al., 2016). Notably, comparative analyses reveal striking geographical variation in sheep resistomes: European flocks are dominated by tetracycline resistance genes (tetW, tetO), while Chinese breeds exhibit elevated β-lactamase abundances (bla_TEM, bla_CTX-M), reflecting local antibiotic usage practices (Zeng et al., 2019). Such divergence highlights the importance of breed- and region-specific resistome profiling, particularly for indigenous varieties such as Tibetan Awang sheep that remain underexplored despite their ecological and agricultural significance.

The Tibetan Awang sheep (Ovis aries), an indigenous breed endemic to the Qinghai-Tibet Plateau, offer a unique model for resistome ecology. Adapted to hypoxic, nutrient-poor environments, Awang sheep possess an enlarged rumen (15–20% greater volume than lowland breeds), which enhances fermentation capacity and supports distinct microbial assemblages (Fan et al., 2021). Traditional grazing practices further expose these sheep to diverse environmental microbial reservoirs, including soil and waterborne ARG carriers (Sabino et al., 2019). Combined with prolonged digestion retention and high microbial density, these factors may impose novel selective pressures on the gut resistome. Moreover, high-altitude ruminants are hypothesized to harbor efflux pump-associated ARGs, potentially linked to detoxification of plant secondary metabolites, a phenomenon previously observed in plateau-adapted livestock (Zeng et al., 2019).

Despite these unique ecological and physiological features, the resistome of Awang sheep remains uncharacterized. Here, we present, to our knowledge, the first metagenomic survey of ARG diversity and abundance in this high-altitude breed. By comparing grazing and captive populations, we provide critical insights into how rearing conditions and environmental exposures shape the gut resistome. This study establishes a baseline framework for AMR surveillance in indigenous Chinese sheep and informs targeted antibiotic stewardship strategies tailored to high-altitude pastoral systems.

The metagenomic sequencing of 100 gut samples generated an exceptionally deep dataset (~1.45 Tb raw data, ~10 billion high-quality reads), providing robust resolution of microbial community dynamics and antibiotic resistance genes (ARGs) distribution in Awang sheep. The high sequencing depth (>14.6 Gb per sample) ensured capture of rare taxa and low-abundance ARGs, which are increasingly recognized as functionally pivotal in gut ecosystems (Ye et al., 2019). Such coverage is particularly important given the complexity of the gut microbiota; minor taxa may disproportionately contribute to digestion, metabolic adaptation, and resistome evolution.

Although alpha diversity (Chao1, Shannon) showed no significant differences between grazing (aw_fm) and captive (aw_qs) sheep, clear beta diversity separation (ANOSIM: r = 0.95, p < 0.001) revealed strong compositional shifts driven by feeding regimen. Firmicutes predominated across both groups, consistent with their established role in fiber degradation and short-chain fatty acid (SCFA) production (Myer et al., 2015). Grazing sheep exhibited higher Bacteroidetes abundance (27.2% vs. 18.1%), likely reflecting specialization in complex polysaccharide metabolism inherent to forage-based diets (Henderson et al., 2015). Conversely, captivity enriched Proteobacteria (13.4%), dominated by Acinetobacter (9.7%), a gene associated with gut dysbiosis and inflammation in livestock (Lima et al., 2024). Elevated Proteobacteria have been linked to impaired suboptimal fiber digestibility, stress, or antibiotic exposure (Shin et al., 2015), suggesting that confinement stressors or dietary additives (e.g., growth promoters) may favor opportunistic taxa, mirroring patterns observed in intensively reared poultry (Campos et al., 2025). LEfSe analysis further reinforced functional divergence, with grazing sheep enriched in Ruminococcus and Bacteroides, key cellulolytic and hemicellulolytic taxa (Biddle et al., 2013). While Acinetobacter emerged as the dominant taxon in captivity. This is particularly concerning, as Acinetobacter is a known reservoir for multidrug-resistant plasmids (Kornelsen and Kumar, 2021). These findings highlight how feeding models shape gut ecology, with potential implications for host health, digestion efficiency, and resistome dynamics.

Resistance mechanisms also diverged between feeding regimens. Target alteration was the dominant mechanism in both groups (50.8% in aw_fm; 34.1% in aw_qs), consistent with its role as a stable evolutionary strategy under low-level selective pressures (Wright, 2011). However, the efflux pump was markedly higher in captive (28.2% vs. 6.5%), suggesting dietary or environmental exposures that favored multidrug resistance strategies. Plant-derived secondary metabolites, such as tannins and flavonoids, known inducers of efflux pump (Blanco et al., 2016), may contribute to this pattern, while horizontal gene transfer (HGT) from environmental bacteria could further amplify efflux-associated resistance (Von Wintersdorff et al., 2016). The inverse relationship between target alteration (a reduction of 16.7% in captivity) and efflux activity (an increase of 21.7%) suggests a potential ecological trade-off; while target alteration offers stable, mutation-driven resistance, efflux provides broader coverage but a higher energetic cost (Andersson and Hughes, 2010). This dynamic implies differential selective pressures—grazing may favor persistent genomic adaptations, whereas captivity imposes intermittent multi-drug pressures favoring efflux.

The predominance of rpoB2 (43.3%), encoding an RNA polymerase β-subunit mutation conferring rifampicin resistance, underscores strong selective pressure. Although rifampicin is rarely used in livestock, cross-resistance to plant-derived RNA polymerase inhibitors such as salicylates or phenolics may explain its high abundance in grazing sheep (Goldstein, 2014). The presence of Bifidobacterium_adolescentis_rpoB (11.2%) further supports this hypothesis, as Bifidobacterium spp. metabolize dietary polyphenols, potentially co-selecting for rpoB mutations (Grimm et al., 2015). However, it is important to note that Bifidobacterium spp. were not directly detected in our samples. This discrepancy suggests that the assignment may not be fully accurate and requires further validation. Future studies should aim to clarify this assignment through additional experimental approaches.

The detection of ugd (10.2%), linked to polymyxin resistance via lipid A modification, likely reflects adaptation to endogenous antimicrobial peptides (e.g., defensins), rather than clinical polymyxin exposure (Olaitan et al., 2014). Efflux-associated genes such as MexK (Pseudomonas aeruginosa) and adeJ/abeM (Acinetobacter baumannii) were enriched in captivity, consistent with anthropogenic antibiotic exposure and subtherapeutic use in intensive farming systems (Zhao et al., 2021). The detection of mupA in 3.2% of grazing sheep is particularly intriguing, considering that mupirocin is primarily used in human medicine (Poovelikunnel et al., 2015). This observation leads us to hypothesize several potential sources for this resistance gene in sheep. One hypothesis is anthropogenic contamination, where environmental exposure to human waste or runoff containing mupirocin residues could introduce the gene into the grazing ecosystem. For instance, studies have shown that human activities can lead to the dissemination of antibiotics in agricultural settings (Kummerer, 2009). Another potential hypothesis is the undocumented off-label use of mupirocin in veterinary medicine, which may occur due to a lack of regulation or oversight, although there is limited evidence to support this practice currently (Fromberg et al., 2017). These hypotheses underscore the need for further investigation, including environmental sampling and interviews with local farmers, to elucidate the pathways through which mupA is introduced and maintained in these populations. Future studies should aim to clarify these mechanisms to better understand the implications for One Health and antimicrobial resistance management.

Resistance profiles revealed contrasting selection pressures: grazing sheep were dominated by rifamycin (35%) and peptide (37.3%) resistance, likely reflecting adaptation to soil-derived antibiotics and host antimicrobial peptides (AMPs) (Donia and Fischbach, 2015). In contrast, captive sheep exhibited broader resistance spectra, with elevated tetracyclines (10.5%), macrolides (9.5%), and fluoroquinolones (7.1%) resistance, indicative of anthropogenic antibiotic use for growth promotion or prophylaxis (Mulchandani et al., 2023). Elevated cephalosporin resistance in captivity is particularly concerning, given the zoonotic potential of plasmid-borne ESBLs (Woerther et al., 2013).

The differential dominance of rpoB2 (48.8% in grazing vs. 25.5% in captivity) further illustrates this dichotomy: grazing environments select for rifamycin-like resistance through natural antibiotic exposure (Wright and Poinar, 2012), while captivity favors efflux-based multidrug resistance. LEfSe analysis confirmed these patterns, with grazing enriched in environmentally acquired ARGs (rpoB2, LlmA_23S_ribosomal, mupA, tet37), while captivity favored efflux-associated determinants (MexK, adeJ, abeM). Notably, the enrichment of adeJ/K was associated with both antibiotic and biocide resistance (Andersson and Hughes, 2014), suggesting cross-selection driven by farm hygiene practices.

Host–microbiome associations revealed species-specific ARG signatures alongside functional convergence. For example, rpoB2 (Nocardia farcinica) and Bifidobacterium_adolescentis_rpoB both confer rifamycin resistance via target alteration, reflecting evolutionary conservation (Spanogiannopoulos et al., 2016). LlmA_23S_ribosomal in Paenibacillus sp. LC231 exemplifies ribosomal target modification, a strategy also observed in clinical pathogens such as Staphylococcus (Leclercq, 2002). Meanwhile, ugd in Escherichia coli K-12 highlights peptide resistance via enzymatic target alteration (Baker et al., 2017). Efflux pumps such as MexK (Pseudomonas aeruginosa) and adeJ, abeM (Acinetobacter baumannii) underscore their role as broad-spectrum determinants in Gram-negative pathogens (Piddock, 2006). The unidentified host of tet37 suggests environmental uncultured reservoirs (Forsberg et al., 2012), underscoring the importance of functional metagenomics to resolve novel resistance origins. We encountered challenges in assigning the tet37 gene to any known organism, which suggests the possibility of an environmental reservoir for this gene. This limitation underscores the complexity of antibiotic resistance gene (ARG) dissemination in microbial communities and highlights gaps in our current understanding of ARG hosts. To address this, future research should use functional metagenomics approaches, which have the potential to uncover novel hosts and provide deeper insights into the ecological roles and dissemination pathways of ARGs (Sommer et al., 2009).

Our analysis suggests that resistome differences among captive sheep are indicative of anthropogenic selective pressure. However, a significant limitation of our study is the absence of antibiotic administration records, which restricts our ability to directly link specific antibiotics to observed ARG patterns. Consequently, our findings are drawn from ecological correlations rather than direct evidence of individual-level antibiotic exposure. This highlights the need for comprehensive records in future studies to better understand the relationship between antibiotic use and resistome composition.

Our research demonstrated that feeding models exert strong ecological pressures to shape both microbial communities and resistomes in Awang sheep. Grazing favors resistance derived from environmental microbial reservoirs and natural antibiotics, while captivity selects for multidrug efflux and clinically relevant ARGs, likely due to antibiotic use and farm management practices. These results have direct implications for One Health frameworks: grazing systems may facilitate environmental ARG infiltration, while confinement amplifies clinically significant resistance, raising zoonotic risks. Future studies integrating cultivation-based approaches and host immune profiling will be essential to validate host–microbiome associations and evaluate potential transmission pathways.