Antibiotic resistance genes in livestock manure
Animal type is the strongest driver of antibiotic resistance genes in livestock manure. Across five species and four production stages, layer litter and pig manure carried the highest resistance gene diversity and abundance. Growth stage shifted the abundance of individual genes such as blaTEM, sul1, tet(Q) and tet(M) without changing the overall resistome structure. Of 944 subtypes found on DNA, only 579 were actively transcribed, and just 29 turned up at both levels.
Why manure resistance is a food chain problem
The global chicken population went from 14.7 billion to 33.1 billion between 2000 and 2020. That is roughly a 120 percent increase in two decades, and chickens are only part of it. Add pigs, sheep and cattle, and the manure and litter produced every day becomes an industrial material flow in its own right. Most of it is land applied as organic fertilizer. That is where the trouble starts.
Veterinary antibiotics are used to treat and prevent disease. A substantial share of each dose leaves the animal unmetabolized, so it arrives in the manure along with bacteria that have already learned to tolerate it. That is why antibiotic resistance genes in manure are so widespread, and why antibiotic resistant bacteria in manure are so difficult to displace. Spread treated manure on farmland and resistance genes move into soil communities, transfer horizontally into new bacterial hosts, and reach the microbiomes of crops growing in that soil. Field work on beef cattle manure has traced the whole sequence: genes enriched in soil, then detected inside the microbiome of lettuce grown on it.
Composting and anaerobic treatment cut the load. Neither eliminates it. And the raw number of genes sitting in a manure pile is not the number that matters most. What matters is how many are switched on inside living bacteria at the moment the manure leaves the farm. A gene present only as dormant or extracellular DNA is a different kind of problem from one being transcribed right now.
That gap between genetic potential and active risk is what an integrated multi-omics study set out to measure across five livestock species.
How the study was designed
Fresh manure came from five commercial-scale farms, one per animal type, each with a herd or flock above 1,000 animals. Animals at different growth stages were housed separately indoors, and antibiotics were used only in limited therapeutic amounts rather than as routine growth promoters. Samples were taken directly after defecation with sterile tools and no bedding material, then frozen at minus 20 degrees Celsius until extraction.
| Parameter | Detail |
| Animal types | Pig, layer (laying hen), sheep, dairy cattle, beef cattle |
| Farms | Five commercial-scale operations, one per animal type |
| Growth stages | Early-life, juvenile, rearing, adult |
| Stage coverage | Pigs, layers, sheep and dairy cattle sampled at four stages; beef cattle at three |
| Biological replicates | Three per animal type and growth stage |
| Targeted quantification | qPCR resistance gene quantification for 12 target genes plus the 16S rRNA gene |
| Composition profiling | Shotgun metagenomics and metatranscriptomics |
| Sequencing output | 80.5 Gb of clean metagenomic data and 20.73 Gb of metatranscriptomic data |
| ARG annotation | The SARG database with BLAST refinement, plus the CARD database for assembled contigs |
| Normalization unit | ARG copies per copy of the 16S rRNA gene |
The 12 genes selected for qPCR cover the resistance classes that show up most often in animal manure: tetracyclines (tet(Q), tet(W), tet(X), tet(M)), sulfonamides (sul1, sul2), beta-lactams (blaTEM), macrolides (erm(B), erm(F)), quinolones (qnrA, qnrB) and polymyxins (mcr-1). Amplification efficiency across all runs landed between 99.2 percent and 109.6 percent, with R squared values above 0.998.
Two design details shape how the results should be read. Growth stage labels were harmonized across species so they could be compared, which means early-life refers to the earliest stage sampled for each animal, such as suckling offspring in mammals and layer chicks within about a week of hatching. Those stages are not physiologically equivalent between species. And because only one farm was sampled per animal type, differences in diet, housing and antimicrobial use at the farm level cannot be separated from species effects.
The resistome at a glance
| Metric | Value |
| Resistance gene subtypes detected on DNA | 944 |
| Resistance gene subtypes actively transcribed | 579 |
| Subtypes detected at both DNA and RNA levels | 29 |
| Dominant resistance classes | Tetracycline, multidrug, macrolide-lincosamide-streptogramin (MLS) |
| Highest DNA-level resistance gene diversity | Layer litter |
| Highest resistance gene transcript abundance | Pig manure |
| Transcriptional diversity coverage | 97.63 percent in pig, 78.62 percent in dairy cattle, 75.84 percent in beef cattle, 67.40 percent in sheep |
| Absolute abundance range across 12 genes | 9.39 million to 2.93 billion gene copies per gram of dry weight |
| Most abundant single gene | erm(B) at 2.93 billion copies per gram of dry weight |
On DNA, the most abundant subtypes were tet(Q), lnu(C), tet(W) and aph(3′)-IIIa. On RNA, the most abundant active subtypes were tet(W), tet(Q), lnu(C), erm(B) and ugd. Aminoglycoside, tetracycline and MLS resistance dominated the active fraction.
Animal type outweighs growth stage in shaping antibiotic resistance genes in livestock manure
Here is the headline result. Almost all of the variation in the livestock manure resistome traces back to which animal produced the manure, not to how old that animal was. The monogastric versus ruminant manure split appears at every level of the analysis.
At the DNA level, principal coordinates analysis of resistance gene profiles separated animal types with the first two axes explaining 74.2 percent and 12.5 percent of total variation. Testing attributed a large and significant share of that separation to animal type (R squared = 0.75, p = 0.001) and found no significant effect of growth stage (R squared = 0.11, p = 0.91). RNA told the same story: 64.8 percent and 17.4 percent on the first two axes, animal type significant (R squared = 0.67, p = 0.002), growth stage not significant (R squared = 0.20, p = 0.62).
The bacterial communities behaved identically. At genus level on DNA, animal type explained a significant share of community variation (R squared = 0.53, p = 0.001) while growth stage did not (R squared = 0.11, p = 0.99). On RNA the pattern held again (R squared = 0.51 for animal type, 0.10 for growth stage).
Firmicutes, Bacteroidota, Pseudomonadota and Actinomycetota dominated across all manures. Enterococcus, Jeotgalibaca and Lactobacillus led at genus level overall, but the detail is species specific. Enterococcus and Lactobacillus were enriched in layer litter. Lactobacillus also turned up enriched in pig manure. Jeotgalibaca was enriched in layer litter and dairy cattle manure. Acinetobacter was mainly a dairy cattle organism, and Gordonia dominated in pig manure.
The animal type and resistome relationship is the single organizing pattern in this dataset. The practical consequence is blunt. Any manure management plan built on the assumption that young animals and adult animals produce meaningfully different resistance profiles is planning around a variable that does not organize the data. Species does.
What “active” means: 944 genes on DNA, 579 on RNA, 29 shared
Metagenomics describes what could happen. Metatranscriptomics describes what is happening. Running both on the same samples opened a gap worth sitting with.
Of 944 resistance gene subtypes detected on DNA, 579 appeared in the transcribed fraction, and only 29 showed up consistently at both levels. The active antibiotic resistome, meaning the genes actually being transcribed rather than merely present as DNA, is the fraction that carries biological risk. Those 29 included tet(W), tet(Q), sul1, sul2, erm(B) and erm(F). Because they are both present and expressed, they are the strongest candidates for tracking environmental risk instead of genetic potential.
The size of the gap moved a lot between animal types. Transcriptional diversity coverage, meaning the share of DNA-detected subtypes that also appeared as transcripts, reached 97.63 percent in pig manure, 78.62 percent in dairy cattle, 75.84 percent in beef cattle and 67.40 percent in sheep manure. Pig manure antibiotic resistance genes were also the most heavily transcribed, and pig manure showed the highest relative abundance of resistance gene transcripts among the animal types with valid read-based transcript profiles.
That qualification carries real weight. Read-based transcript profiling did not produce reliable resistance gene profiles for layer samples, so layer litter was left out of the RNA-level comparison and could only be assessed through assembly-based annotation. The two approaches are not directly comparable, and any cross-species claim about layer litter expression levels inherits that limitation.
Layer litter carries the heaviest resistance gene load
Layer litter stands out on nearly every DNA-level diversity metric. The layer litter antibiotic resistance profile is the heaviest of the five animal types.
Subtype diversity measured with the Shannon index on assembled DNA contigs was highest in layer litter at 5.414 to 5.885, with a Chao1 index of 532.789 to 1050.835. Pig manure ranked second on Shannon at 5.046 to 5.509. The three ruminant species clustered together at 5.037 to 5.420. On RNA transcripts, layer samples again posted the highest Shannon index at 5.65 to 5.8 for subtype diversity, although richness estimates did not differ significantly among animal types.
Targeted qPCR confirms this is not a sequencing depth artifact. Absolute abundances of blaTEM, erm(B), erm(F), qnrB, sul1, sul2 and tet(M) were significantly higher in layer litter than in the other animal types. Normalized to 16S rRNA gene abundance, layer litter carried the highest erm(B) gene abundance at 0.699 copies per 16S rRNA gene, followed by erm(F) at 0.631, sul2 at 0.220, sul1 at 0.0696 and tet(M) at 0.0387.
Sheep manure went the other way for one gene family. Absolute and relative abundances of tet(Q), tet(W) and tet(X) were highest in sheep, with tet(Q) at 0.0992, tet(X) at 0.0772 and tet(W) at 0.0365 copies per 16S rRNA gene. Across every tetracycline gene measured, tet(X) reached the highest absolute concentration at 353 million copies per gram of dry weight.
Why would layer litter carry the heaviest load? Intensive indoor housing creates repeated opportunities for gut colonization by resistance-carrying taxa and constant exchange with the barn environment. Cages and feathers are already known to act as environmental reservoirs that feed the resistance gene pool in layer systems. Lactobacillus and Enterococcus were both enriched and transcriptionally active in layer litter, which ties community composition straight to the gene profile.
Growth stage moves individual genes but not the overall structure
Growth stage failing to change overall resistome composition does not make it irrelevant. Targeted qPCR shows it moves individual genes, sometimes hard. Any growth stage antibiotic resistance signal is gene specific rather than structural.
Absolute and relative resistance gene abundances were highest during the early-life and juvenile stages. Testing identified significant growth stage effects on blaTEM, erm(B), qnrA and sul1. Pairwise comparison showed that relative abundances of blaTEM and tet(Q) were significantly higher at the juvenile stage than at other stages, while tet(M) was significantly higher at the early-life stage.
The fold changes from early-life to juvenile are the most striking numbers in the dataset. In pig manure, blaTEM gene abundance rose roughly 40-fold and sul1 gene abundance roughly 71-fold over that interval. In sheep manure, tet(Q) rose about 10.5-fold and tet(W) about 7.0-fold across the same transition.
The pattern lines up with how antibiotics are actually used. Tetracyclines are commonly given to pigs to treat or prevent respiratory disease during the post-weaning and fattening windows, which is exactly when tetracycline and aminoglycoside resistance genes peak. In layer flocks, multidrug and tetracycline resistance genes are enriched during the adult stage. That reflects microbiome succession, persistent early-life effects and cumulative exposure from production-scale conditions rather than a single medication event.
Cattle resistomes are the most stable of the five. Manure from dairy and beef production shifts comparatively little across grazing versus confined feeding, diet composition and manure handling differences.
The counterintuitive part: a DNA-only survey would misjudge the risk
Most resistance gene surveillance stops at DNA. Comparing the DNA and RNA layers shows what that shortcut costs.
Count only the genes present in manure and layer litter looks like the single largest risk of the five animal types, with all 944 detected subtypes weighted equally. Add the transcriptional layer and the ranking changes in two ways. Layer litter still carries the highest gene diversity, but it cannot be placed on the expression axis with the same confidence, because reliable read-based transcript profiles were not obtained for those samples. Meanwhile pig manure, second on DNA diversity, turns out to be the animal type where the largest share of detected genes is actually being expressed, at 97.63 percent coverage.
A second surprise sits in the relationship between microbial diversity and resistance diversity. Sheep manure carried the highest bacterial community diversity of any animal type, with a Shannon index of 5.434 to 6.388 at the DNA read level. That did not translate into the highest resistance gene diversity, which belonged to layer litter. High microbial diversity and high resistance gene diversity are separate properties. Ecological shifts such as weaning and diet-driven community restructuring can reshape the resistome without moving overall bacterial richness at all.
The conclusion this dataset supports is narrow and worth stating plainly: metagenomics alone cannot assess functional antimicrobial resistance risk. A gene catalog tells you what the manure could do. Transcription tells you what it is doing. Risk assessment that needs the second answer has to measure it.
Microbiome structure tracks the resistome at both levels
The study also tested whether resistance gene profiles follow bacterial community structure. They do, closely.
Procrustes analysis, which measures how well two ordinations can be aligned, produced a strong fit at both levels. At the DNA level the residual sum of squares was 0.125 with a correlation of 0.935 and p = 0.001. At the RNA level the residual was 0.312 with a correlation of 0.829 and p = 0.001.
The distance between those two residual values is the interesting part. The weaker fit at RNA level means transcriptional resistome profiles diverge further from community structure than DNA profiles do. That makes biological sense. DNA-level presence is largely governed by which taxa are there. RNA-level expression also depends on microbial physiology, host metabolism and local selective pressure. Presence is fairly predictable. Expression is conditional.
This has a direct implication for monitoring design. If the goal is to predict which resistance genes will be present, bacterial community composition is a reasonable proxy. If the goal is to predict which genes will be active, community composition alone will overstate how well it works.
What this means in practice
- Sort sampling by animal type before growth stage when the objective is to characterize a farm’s resistance gene profile. Species explains far more of the variance than age, so manure management antibiotic resistance priorities should follow species first.
- Treat layer litter as the highest-priority stream for resistance gene management. It carries the highest gene diversity plus the highest absolute and relative abundances of blaTEM, erm(B), sul1, sul2 and tet(M).
- Watch the early-life to juvenile transition in pigs and sheep. This is where blaTEM, sul1, tet(Q) and tet(W) jump the most, so it is the window where manure handling decisions pay off most.
- Do not read a gene catalog as a risk score. Twenty-nine subtypes were both present and transcribed. Those are the ones worth tracking with targeted qPCR.
- Use community composition as a screening tool, then confirm with targeted expression assays, rather than running full metatranscriptomics on every batch.
- Segment manure streams by species before treatment. A single composting protocol applied across a mixed-species operation treats very different inputs as if they were the same material.
Limitations worth knowing
Only one commercial farm was sampled per animal type, so species effects and farm-level differences in diet, housing, antimicrobial exposure, management practice and manure handling are confounded. The results cannot separate them.
Growth stages were harmonized across species for comparability but are not physiologically equivalent, and the design was unbalanced because beef cattle were not sampled at all four stages. Biological replicates were few and management and antimicrobial use records were incomplete, which limits statistical power and generalization.
Reliable read-based transcript profiles were unavailable for layer samples. Assembly-based annotation did detect transcriptionally represented genes in layers, but those estimates are not directly comparable with read-based estimates from the other animal types.
These are observational findings from one sampling campaign. They describe associations inside this dataset. They are not a validated predictive model, and no intervention was tested, so the effectiveness of any manure treatment or feed management strategy remains unproven here.
Bottom line
In livestock manure, the species of animal matters more than its age for determining which antibiotic resistance genes are present. Growth stage matters for how abundant individual genes are, with the early-life and juvenile windows carrying the heaviest loads and the largest jumps. And the gap between 944 genes detected and 579 genes transcribed is a reminder that presence and activity are different measurements. Only the second one describes an active biological risk.
FAQ
Which antibiotic resistance genes are most abundant in livestock manure?
The most abundant subtypes on DNA were tet(Q), lnu(C), tet(W) and aph(3′)-IIIa. On RNA, the most abundant active subtypes were tet(W), tet(Q), lnu(C), erm(B) and ugd. By targeted quantification, erm(B) reached the highest absolute abundance at 2.93 billion copies per gram of dry weight, and tet(X) led the tetracycline genes at 353 million copies per gram.
Does growth stage change antibiotic resistance genes in livestock manure?
Growth stage does not significantly change the overall composition of the manure resistome, but it does change the abundance of individual genes. Gene abundances were highest during the early-life and juvenile stages. In pig manure, blaTEM rose about 40-fold and sul1 about 71-fold between early-life and juvenile stages. In sheep manure, tet(Q) rose about 10.5-fold and tet(W) about 7.0-fold.
Why does layer litter carry so many antibiotic resistance genes?
Layer litter showed the highest resistance gene diversity of the five animal types, with a subtype Shannon index of 5.414 to 5.885, and significantly higher abundances of blaTEM, erm(B), erm(F), qnrB, sul1, sul2 and tet(M). Intensive indoor housing creates repeated colonization opportunities and constant exchange with barn surfaces, cages and feathers, which act as environmental reservoirs that feed the resistance gene pool.
What is the difference between the DNA-level and RNA-level resistome?
The DNA-level resistome describes which resistance genes are present in the sample, including dormant or extracellular DNA from dead cells. The RNA-level resistome describes which of those genes are actively transcribed inside living bacteria. In this dataset, 944 subtypes appeared on DNA and 579 appeared as transcripts, with only 29 subtypes detected at both levels.
How many antibiotic resistance genes are actually active in manure?
Of 944 resistance gene subtypes detected on DNA, 579 were actively transcribed, and 29 were consistently detected at both levels, including tet(W), tet(Q), sul1, sul2, erm(B) and erm(F). The share of detected genes that were also expressed varied by species, reaching 97.63 percent in pig manure, 78.62 percent in dairy cattle and 67.40 percent in sheep.
Which animal manure has the highest antibiotic resistance gene diversity?
Layer litter had the highest resistance gene diversity of the five animal types, with a Shannon index of 5.414 to 5.885 and a Chao1 index of 532.789 to 1050.835 on assembled DNA contigs. Pig manure ranked second at 5.046 to 5.509. Sheep, dairy cattle and beef cattle clustered together at 5.037 to 5.420.
Do monogastric animals carry more resistance genes than ruminants?
Yes, in this dataset. Pig and layer samples showed higher resistance gene abundance and diversity than sheep, dairy cattle and beef cattle. Layer litter reached the highest abundance of several genes after 16S normalization, including erm(B) at 0.699 and erm(F) at 0.631 copies per 16S rRNA gene. Ruminants carried lower richness and distinct communities, with sheep highest only for tet(Q), tet(W) and tet(X).
What methods are used to detect active antibiotic resistance genes?
Two methods work together. Metagenomic sequencing identifies which resistance genes are present by annotating DNA reads against a resistance gene reference database. Metatranscriptomic sequencing identifies which genes are being expressed by sequencing cDNA after ribosomal RNA depletion. Targeted qPCR then quantifies specific genes against the 16S rRNA gene to give comparable abundance values across samples.
Can manure treatment reduce antibiotic resistance genes?
Treatment reduces the load but does not eliminate it. Composting and anaerobic digestion lower resistance gene abundance, and transcriptional profiling shows composting can also reduce expression of the genes that remain. The prioritization signal here is species and timing: layer litter and manure from the early-life to juvenile window carry the highest gene abundance, so those streams deserve the most aggressive treatment.
How much do blaTEM and sul1 rise between early-life and juvenile stages?
In pig manure, blaTEM increased roughly 40-fold and sul1 increased roughly 71-fold between the early-life and juvenile stages. Those were the largest fold changes recorded among qPCR targets across the five animal types. They coincide with post-weaning and fattening medication windows, when tetracyclines and other antibiotics are most commonly used in pig production.
Is antibiotic resistance in livestock manure a human health risk?
The pathway is real and documented. Manure applied to farmland enriches resistance genes in soil and can transfer them into crop-associated microbiomes, including lettuce, through horizontal gene transfer. The genes most relevant to risk are not all 944 detected subtypes but the 29 found to be both present and actively transcribed, because transcription indicates a working resistance mechanism rather than dormant DNA.
What are the limitations of this type of resistome study?
Three limits matter. Only one farm was sampled per animal type, so species effects are confounded with differences in diet, housing and antimicrobial use. Growth stages were harmonized across species, are not physiologically equivalent, and beef cattle were not sampled at all four stages. Reliable read-based transcript profiles were also unavailable for layer samples, limiting cross-species comparison of expression.
