Farm mice carry the fingerprints of nearby antibiotics

If you want to know where antibiotic resistance lives, don’t start in a hospital ward—start in a barn. A new field study asked a simple, unsettling question: which matters more for the drug-resistance genes in a wild mouse’s gut—the mouse’s own biology, or the farm landscape it scurries through?

The team behind the work, published in Nature Communications by Morgane Gicquel and colleagues, went looking for a measurable answer. They sifted through the gut microbiomes of 875 wild house mice caught on German farms and matched what they found to how those farms and their surroundings are used. Then they compared the mice’s resistance gene profiles to what turns up in cattle, pig and chicken manure in public genomic datasets. The headline number they brought back: about half of the resistance genes cataloged from livestock were also present in the wild mice.

What they set out to test—without the usual hedging

The researchers wanted to know the relative weight of environment versus host when it comes to antibiotic resistance in wildlife. In plain terms: are wild mice carrying whatever resistance genes their own bodies and diets select for, or are they picking up a signature of the farms they live alongside?

To probe that, they examined how many and which antibiotic resistance genes (ARGs) turned up in each mouse’s gut microbiome, and how that portfolio changed with local land use, specific farming practices and the density of livestock nearby—cattle, pigs and poultry—alongside the mouse’s own traits (sex, physical condition) and climate variables.

Who did it and where it ran

The study comes from a collaboration including Dr. Víctor Hugo Jarquín-Díaz of the Max Delbrück Center; Professor Emanuel Heitlinger, who led the work at Humboldt-Universität zu Berlin and the Leibniz Institute for Zoo and Wildlife Research (Leibniz-IZW) and is now associated with the Federal State Agency for Consumer & Health Protection Rhineland-Palatinate; and Professor Stephanie Kramer-Schadt of the Leibniz-IZW and Technische Universität Berlin.

Their paper—“Farming practices exert selection pressures on the resistome of natural populations of house mice”—appears in Nature Communications (2026), and the findings were summarized by Phys.org, which quoted the team and reported the key statistics and comparisons.

How they measured it—875 mice, farm by farm

The method has two steps. First, the scientists analyzed the genomes of gut microorganisms from 875 wild house mice (Mus musculus) caught on farms across Germany. They screened those microbiome genomes for antibiotic resistance genes—ARGs—to build each mouse’s resistance profile. Then they ran statistical analyses linking those profiles to a menu of variables: land use and farming practices on and around each farm; livestock density for cattle, pigs and poultry; the mouse’s sex and physical condition; and climatic factors. The stated goal was to explain which factors shape which ARGs show up, and how many.

Second, they compared the mice’s resistance profiles to ARGs cataloged in the manure of farm animals—cattle, pigs, chickens—using genomic data from other projects that are publicly available. This cross-reference asks, in effect: how much overlap is there between the resistome—the full set of resistance genes—in local livestock waste and in the guts of nearby wild mice?

What turned up in the data—numbers, not adjectives

Two numeric statements define the result. One: about 50% of the resistance genes cataloged from cattle, pigs and chickens were also present in the wild mice sampled. The team expected “some overlap,” but this level surprised them. Two: environmental variables and the intensity of livestock farming explained more of the variation in the mice’s resistomes than the mice’s own characteristics did.

That second claim comes with a precise comparison: “The way agricultural land in the immediate vicinity of the farm is used, and the resulting direct and indirect contact with farm animals, explains three times as much about which ARGs are detectable in the mouse as the mouse’s sex or body condition.” In addition, the authors report that pig farming density is strongly associated with resistance genes tied to antibiotic classes widely used in veterinary and farming practices, naming sulfonamides, tetracyclines and beta-lactams.

What that means in real life on a farm

All else equal, a mouse living where pigs are densely farmed is more likely to carry ARGs associated with the drugs those pigs see most often. And when the land right around a farm is used in ways that bring wildlife into direct or indirect contact with livestock—fields, feed stores, waste areas—that context is a stronger predictor of which ARGs show up in a mouse than whether that mouse is male or female, fat or thin.

The authors’ broader framing is blunt: there are “many ecological bridges between humans, farm animals and wildlife,” and in the case of antibiotic-resistant bacteria, those bridges are evidently in use. The team argues that resistance monitoring should expand beyond clinical and livestock settings, because wild animals and the ecosystems they move through may be sizable, unmonitored reservoirs of antibiotic resistance.

Item	Detail
Study species and sample size	Wild house mice (Mus musculus), n = 875
Where mice were sampled	Farms across Germany
Primary measurement	ARGs in gut microbiome genomes of mice
Predictors tested	Land use; farming practices; livestock density (cattle, pigs, poultry); mouse sex; mouse condition; climatic variables
Cross-system comparison	Mice ARG profiles vs. ARGs in cattle, pig, chicken manure from public genomic datasets
Key overlap result	~50% of livestock resistance genes also present in wild mice
Relative influence	Local agricultural land use and contact with farm animals explained ~3× more about detectable ARGs than mouse sex or condition
Specific association	Pig farming density strongly associated with ARGs linked to sulfonamides, tetracyclines, beta-lactams

Method matters: what’s in, what isn’t

This is not a hospital chart review; it’s an ecological genetics survey. The core data are ARGs extracted from the gut microbiome genomes of hundreds of free-living mice, each tied to a specific farm context. The analysis is explicitly statistical—linking ARG presence to environmental and host variables—and comparative, by setting the mice’s resistomes against livestock-manure gene catalogs built by other projects.

Important practical details are not given in the summary: which sequencing platforms were used, how ARGs were annotated, which statistical models were fitted, and over what time period the mice were sampled. The authors’ own phrasing, as reported, emphasizes explained differences (for example, “three times as much”) and named associations (pig density with sulfonamide/tetracycline/beta-lactam-linked ARGs), rather than specific p-values or error bars.

What the numbers do—and do not—show

“Around 50%” overlap between livestock and mouse ARG catalogs tells you there is a large, concrete intersection between the resistomes of farm animals and wildlife on German farms. It does not claim that any specific gene moved in a particular direction or on a particular day. The threefold explanatory power of local land use over mouse sex or condition means environmental context dominated the modeled variance in which ARGs appeared; it does not convert to a single risk number for any one mouse.

The association between pig density and ARGs linked to widely used veterinary antibiotics points to a match between what is used on farms and what turns up in wildlife. It is, as presented, an association: a direct causal mechanism is not tested here. What is shown, decisively, is that environmental and farming variables “have a greater influence on the specific resistance profile in the gut microbiome of a wild farm mouse than the mouse’s own characteristics,” in the study’s words as reported.

Limits and honest caveats

Several boundaries are explicit or implied in the source. First, the work is observational and statistical, not an experiment that manipulates antibiotic use or wildlife exposure. It links patterns; it does not trace individual transmission events or identify the direction of gene flow. Second, some important methods details are not provided in the summary: the sequencing approach, the exact statistical models, and the sampling timeline. Without those, readers can’t audit platform-specific biases or the temporal stability of the patterns reported here.

Third, the cross-system comparison relies on publicly available genomic data for manure from cattle, pigs and chickens compiled by other projects. That is a strength—it leverages large datasets—but the summary does not detail how those datasets were harmonized with the mouse data. Finally, the geography is specific: farms in Germany. The authors’ language invites generalization to “ecosystems heavily used and shaped by humans,” but demonstrating that in other regions would require similar fieldwork elsewhere.

Why it matters for policy and monitoring

The authors argue that antibiotic resistance is “not an isolated medical problem, but a systemic ecological phenomenon.” In their framing, the practical upshot is straightforward: expand resistance monitoring beyond clinics and barns, into wildlife and the environments that connect them. Mice are not patients, but they are neighbors of livestock and, as this study shows, they carry a substantial share of the same resistance genes.

Mapping those “ecological bridges” the team talks about—where land use and farm practices bring animals into direct and indirect contact—would let agencies place surveillance where it’s most informative. If environmental context explains several times more about wild resistomes than host traits do, then surveillance designs should start with land-use and livestock-density layers, and only then stratify by the animal’s sex or condition.

The next obvious tests

Two follow-ons suggest themselves from the authors’ own emphasis. First, build the “map” they call for: a spatially explicit picture of how land use around farms structures the resistomes of wildlife, using the same measures across regions. Second, test the named associations more finely—pig density versus ARGs tied to sulfonamides, tetracyclines and beta-lactams—by asking whether changes in those farming practices over time track changes in nearby wildlife resistomes.

Even before such work, this study shifts the burden of proof. If about half of livestock resistance genes show up in wild mice on farms, and if local land use predicts mice’s resistomes three times better than sex or body condition, then any resistance policy that ignores wildlife and the spaces that connect barns to fields is missing half the story.