Intensified livestock farming increases antibiotic resistance genotypes and phenotypes in animal feces

The study addresses how intensification of livestock farming alters the reservoir of antibiotic resistance in animal feces, a key vector for environmental dissemination under the One Health framework. With projected increases in antibiotic use driven by growing animal populations and shifts toward intensive production, there is concern that higher selective pressures and incomplete metabolism of antibiotics will elevate resistance in fecal microbiomes. The research investigates whether and how resistance genotypes (ARGs and MGEs) and phenotypes increase with farming intensity, their association with bacterial community composition and MGEs, and the role of potential pathogens as carriers, by comparing free-range, small-scale, and intensive farms across multiple animal species.

Prior work shows feces from intensive operations harbor ARG abundances orders of magnitude above background soils, with dairy cattle feces carrying hundreds of ARGs and commercial poultry feces enriched in tetracycline and aminoglycoside ARGs. Comparisons among conventional, antibiotic-free, and organic systems indicate lower resistance in isolates from antibiotic-free/organic farms than conventional ones, though environmental exposure and legacy use may confound outcomes. In cattle feedlots, organic management correlated with much lower ARG abundances than therapeutic-antibiotic use. Intensification is expected to increase antibiotic use and excretion via feces, while MGEs facilitate horizontal transfer and co-selection of multidrug resistance. However, how resistance reservoirs in feces vary specifically along intensification gradients and the coupling between genotype and phenotype across entire fecal microbiomes has been underexplored.

Study area and sampling: Eight farms on the southwestern rim of the Qinghai-Tibet Plateau (Southwest China) were surveyed: four free-range alpine pastures (yak, sheep, pig, horse), three small-scale captive farms (<500 animals; yak, sheep, pig), and one intensive pig farm (>10,000 animals). Fresh feces from six healthy adult animals per farm (n=48) were collected in May 2020, transported on dry ice, and split for -80°C storage (physicochemical analyses, DNA) or 4°C storage (isolation). A pooled agricultural soil sample (Hani Terrace paddy soil) served as a phenotypic reference. Physicochemical analyses: Measured pH and organic matter; quantified Cu, Zn, Pb, Cd, Cr; and 34 antibiotics across 10 classes using HPLC-MS/MS (LCMS-8040), following established sample preparation and gradient methods, with rapid processing for beta-lactams. High-throughput qPCR: Extracted DNA (PowerFecal kit, cleaned with DNeasy PowerClean). Quantified 317 ARGs (10 classes), 57 MGEs, and 16S rRNA on a SmartChip Real-time PCR platform. Relative abundances calculated via ΔΔCt normalized to 16S rRNA; Ct=31 used as detection limit; triplicate technical replicates required concordant amplification. Absolute abundances derived by normalizing relative abundances to absolute 16S rRNA gene copies (quantified via gradient dilution qPCR). 16S rRNA gene amplicon sequencing: V3–V4 region (341F/805R) sequenced (MiSeq 2x300). Reads (mean ~53k/sample) clustered at 97% identity (UCLUST), taxonomically assigned (RDP), and rarefied. Potential pathogens identified by matching OTUs (>99% identity, E-value <1e-10) against a curated database of 538 pathogenic bacterial species. Microbiome phenotypes inferred with BugBase. Analyses performed in QIIME 2. Shotgun metagenomics: 24 fecal DNA samples (3 per farm) sequenced (NovaSeq 6000, PE151). Quality-trimmed reads assembled with MEGAHIT; ORFs predicted (MetaGeneMark). ARGs annotated against CARD; MGEs against customized databases (NCBI RefSeq plasmids, ISfinder, INTEGRALL) using DIAMOND BLASTx (≥80% identity; aa length >25). Abundances reported as TPM. Co-localization assessed by screening ARG-containing contigs for MGE hits; ARG host taxa inferred by mapping contigs to NCBI NR and cross-referencing to the pathogen list. Bacterial isolation and identification: Serial dilutions of feces/soil plated on Nutrient Agar; colonies selected over 7 days at 30°C to maximize phylogenetic diversity. Pure isolates maintained on LB slants. 16S rRNA genes amplified (27F/1492R) and identified via EzBioCloud; full-length sequencing for low-confidence matches. Resistance phenotype profiling: Susceptibility of 1324 identified isolates to 34 antibiotics assessed using standard disc diffusion (Mueller-Hinton agar); inhibition zone diameters (IZD) recorded (IZD=0 if growth within zone). CLSI 2011 disc potencies used; E. coli ATCC 25922 and P. aeruginosa ATCC 27853 as QC strains. Statistical analyses: Variation partitioning analysis (vegan) to apportion ARG variance among abiotic factors (basic/nutritional properties, antibiotics, heavy metals) and bacterial community composition. Random forest models to predict ARGs from dominant genera (top 100) and from putative pathogens. Multidrug resistance assessed via co-occurrence of ARGs and phenotypic resistances using Spearman correlations and chi-square (Cramer's V); ratios of between-class to within-class connectivity/significance computed to normalize multidrug signals. Phylogenetic signal of resistance phenotypes evaluated with Blomberg’s K (picante). Nonmetric multidimensional scaling and other ordinations used for community analyses; nonparametric tests (Kruskal–Wallis, Wilcoxon) for group comparisons.

- Chemical drivers: Intensive farms had significantly higher micro-/meso-elements and heavy metals in feces; Cu and Cd elevated (p<0.001); Zn extremely high in intensive pig feces (10.49 mg g−1). Of 34 antibiotics, ~10 detected per sample; higher concentrations in captive systems. Oxytetracycline reached 33.92 mg kg−1 in intensive pig feces vs 0.119 mg kg−1 in free-range. - ARG diversity and abundance: High-throughput qPCR detected 207/317 ARGs and 42/57 MGEs across samples. Free-range yak feces had the fewest ARGs (mean 34), intensive pig feces the most (137). Mean relative ARG abundance in intensive pig feces was 46.0× and 10.1× higher than free-range and small-scale farms (p<0.001). ARG classes particularly enriched with intensification: aminoglycoside, MLSB, multidrug, sulfonamide, tetracycline. - MGE profiles and ARG–MGE relationship: More MGEs detected with intensification (e.g., 28 MGEs in intensive pig vs 15 in small-scale pig). Transposases showed the largest relative abundance differences. ARG and MGE relative abundances were strongly correlated (R=0.826, p<0.001), with MGEs increasing proportionally more than ARGs along intensification gradients. Absolute abundance trends matched relative trends. - Community structure and drivers: Bacterial communities (16S) clustered by farming intensity; Firmicutes and Bacteroidetes dominated, with genus-level shifts by animal type. BugBase predicted more MGE-carrying bacteria in captive vs free-range samples (p<0.001). Variation partitioning: antibiotics (22.8%), basic/nutritional properties (15.3%), heavy metals (5.64%) explained ARG variation; adding community composition explained an additional 11.3%, leaving 2.31% unexplained. - Pathogens as ARG carriers: Proportion of pathogens in communities increased with intensification. Random forest models predicted more ARGs from pathogen taxa (27 ARGs) than from dominant genera (14), notably multidrug ARGs. Escherichia coli, Bacteroides spp., and Clostridium perfringens were key predictors/hosts. - Metagenomics confirmation: ARG and MGE TPMs were higher in intensive farm feces. Co-localization of ARGs with MGEs on contigs increased from 17.52% (free-range) to 27.38% (captive); in intensive pig feces, 38.6% of ARG contigs had MGEs. Frequent combinations included multiple transposases with aminoglycoside, sulfonamide, and tetracycline ARGs. Among annotated ARG-carrying contigs, major hosts were Escherichia (6.60%), Clostridium (4.20%), Bacteroides (3.64%). Of ARG-carrying species-level assignments, potential pathogen proportions were ~81.2% (intensive pig), 69.6% (small-scale yak), vs 31.6% (free-range). - Phenotypic resistance: 1324 isolates (Proteobacteria, Firmicutes, Actinobacteria, Bacteroidetes) showed higher resistance (lower mean IZD) in captive feces; soils had the lowest resistance. Highest phenotypic resistance to sulfonamides, several tetracyclines, and aminoglycosides. Mean number of antibiotic classes resisted per isolate: soil 4.0, free-range 5.6, captive 6.4. Multi-antibiotic resistance index: 0.30 (soil), 0.43 (free-range), 0.53 (small-scale), 0.58 (intensive). MDR prevalence 84.6–100%; XDR 8.33–47.0% (highest in small-scale yak). Opportunistic pathogens (Acinetobacter, Bacillus, Enterococcus) often had greater resistance; Pseudomonas and Enterobacteriaceae less consistently elevated. - Multidrug resistance co-occurrence: ARG co-occurrence pairs were far fewer in free-range vs captive (χ² significant pairs: 779 vs 3,035; Spearman average connectivity: 6.770 vs 33.58). Normalized between-class co-occurrence (Ratio_multi) was higher in captive vs free-range (χ²: 5.620 vs 2.885; Spearman: 5.398 vs 4.133). Phenotypic resistance networks also showed more significant pairs (216 captive vs 128 soil, 130 free-range), higher connectivity (avg 7.294 captive vs 5.412 soil, 6.529 free-range), and higher between-class co-occurrence ratios (0.714 captive vs 0.333 soil, 0.300 free-range). - Phylogeny–phenotype decoupling: Resistance phenotypes clustered strongly by antibiotic class and weakly by phylogeny. Blomberg’s K indicated stronger phylogenetic signal in soil and lowest in captive feces, suggesting intensified farming and elevated MGEs decouple resistance phenotypes from bacterial phylogeny.

The findings demonstrate that as livestock farming intensifies, fecal microbiomes exhibit expanded resistomes and mobilomes, with elevated antibiotic and metal concentrations providing selective pressure. The strong correlation between ARGs and MGEs, increased contig-level co-localization, and higher predicted and observed roles of pathogens (e.g., Escherichia, Bacteroides, Clostridium) as ARG carriers indicate enhanced horizontal gene transfer potential, underpinning multidrug resistance development. Phenotypic assays corroborate genotypic patterns, particularly for aminoglycosides, tetracyclines, and sulfonamides—antibiotic classes commonly used in animal production—showing increased multidrug resistance and co-occurrence across antibiotic classes in captive systems. The reduced phylogenetic signal of resistance traits in captive fecal isolates implies decoupling of resistance from lineage due to mobile element-mediated gene flow under strong selection. Collectively, the results address the research question by linking farming intensification to increased resistance genotypes and phenotypes and by elucidating mechanisms (MGE-mediated transfer, co-selection) and carriers (pathogens), highlighting implications for environmental dissemination and One Health risks.

This study provides an integrated genotype-to-phenotype assessment showing that intensification of livestock farming increases the diversity and abundance of ARGs and MGEs in animal feces, enhances multidrug resistance, strengthens ARG–MGE co-localization, increases pathogen carriage of resistance, and decouples resistance phenotypes from bacterial phylogeny. These outcomes suggest greater potential for environmental spread of resistance from intensively managed farms. Policy and practice should prioritize preserving lower-intensity practices where feasible, improving treatment and disposal of animal wastes from intensive farms, and implementing One Health-aligned prevention strategies. Future research should further resolve transmission pathways from feces to soils, crops, and humans; quantify fitness costs and persistence of resistant isolates; monitor wildlife–livestock interfaces; and evaluate interventions that reduce antibiotic and MGE selection pressures in animal production.

Culture-based isolation captures only a subset of fecal microbiota; notably, despite metagenomic indications of Escherichia coli as a key ARG host, few E. coli strains were isolated, limiting phenotypic assessment of that taxon. Many isolates from intensive farms grew slowly with low viability, suggesting potential fitness costs of resistance but requiring more data to confirm. Free-range animals were not antibiotic-free, complicating binary contrasts between management systems. While co-localization of ARGs and MGEs suggests potential horizontal transfer, direct transfer events were not demonstrated.