Cat Owners and DNA Metabarcoding Reveal What Pet Cats Really Eat

The domestic cat is one of the world’s most successful predators, yet it lives a double life: beloved companion by day, stealthy hunter whenever a door swings open. Understanding exactly how much wildlife owned cats actually eat has long frustrated scientists, because traditional monitoring methods—from cat-mounted cameras to stomach analysis—are invasive, expensive, or unreliable. Now, a new pilot study from the University of California, Davis, published in Ecology and Evolution, shows that a combination of citizen science and DNA metabarcoding can turn ordinary cat owners into data collectors, scooping their pets’ litter boxes in the name of conservation science.

The research team, led by Hee Jin Chung of the Mammalian Ecology and Conservation Unit at UC Davis’s Veterinary Genetics Laboratory, recruited owners of indoor–outdoor cats in Davis and Sacramento, California. Participants were given simple kits containing ethanol-filled tubes, nitrile gloves, and a printed log, and were asked to collect five of their cat’s fecal samples over several weeks, along with a sample of whatever food the cat was being fed. Between November 8 and December 21, 2023, the team received 38 scats from eight neutered house cats—three males and five females—whose samples were, for anonymity, labeled Cat A through Cat H. Most samples came straight from litter boxes and were collected within hours of defecation, making them unusually fresh for molecular work.

The technical core of the study was DNA metabarcoding, a sequencing approach that identifies the mixture of DNA left behind in feces. The researchers amplified two mitochondrial markers: the 12S rRNA gene to detect vertebrate prey and the 16S rRNA gene to detect invertebrates. Because cat DNA overwhelmingly dominates cat feces, the team designed a custom blocking oligonucleotide—a short synthetic molecule carrying a three-carbon spacer—that binds to domestic cat sequences and stalls their amplification, allowing dietary DNA to be recovered more easily. Sequencing was performed on an Illumina platform, and the resulting reads were processed with cutadapt and DADA2, which identify distinct amplicon sequence variants, before taxa were assigned using BLAST searches against custom and NCBI reference libraries. Samples below strict read-count and identity thresholds were discarded to guard against contamination.

The results were striking, and somewhat humbling for anyone who assumes their cat is quietly devastating the neighborhood. Of 32 analyzable scat samples, roughly a third contained no vertebrate DNA at all except that of the cat itself. Most of the remainder revealed only anthropogenic food items—chicken, pig, turkey, and fish from the drum, hake, and herring families—matching the provisioned pet foods owners submitted for comparison. In total, only a single scat, produced by Cat H, contained wild vertebrate prey: the California vole, Microtus californicus, a common rodent in the region. The cat that produced it had not been seen with any prey by its owner in the preceding 24 hours.

Even more revealing were the three cases where owners did witness their cats carrying prey—a live house mouse caught indoors, a small unidentified rodent brought to the door, and a lizard caught in front of the home. In every instance the prey was alive and intact, and in every instance the corresponding scat, produced within 24 hours, contained no wild prey DNA whatsoever. The findings suggest that domestic cats frequently hunt without consuming what they catch, consistent with earlier camera-collar studies showing that satiated cats hunt for enrichment, practice, or caching rather than nutrition. They also imply that significant prey consumption is needed before prey DNA becomes detectable in feces, meaning molecular scat analysis and owner reports capture different facets of the same hunting behavior—and should be used together.

The comparison between provisioned food and scat also yielded a methodological surprise. Chicken and pig DNA appeared in all four collected food samples, as expected, but recovery of those ingredients in the corresponding scats varied widely. Chicken, present in nearly every relevant food, showed up in 14 of 15 expected scat samples, making it the most digestible and detectable ingredient. By contrast, the two prescription diets—a hydrolyzed-protein formula and a kidney-care wet food—showed the poorest post-digestion recovery, likely because heavy industrial processing degrades DNA before the food even reaches the bowl. Notably, several scats from cats eating animal-based diets contained no animal DNA other than cat DNA, a finding with implications beyond cats: urban wildlife studies routinely identify pet food as an anthropogenic food source in carnivore scats, and this study suggests such DNA may often be too degraded to detect.

The team also tested a cost-saving shortcut that could matter for future studies: pooling. Because DNA extraction and sequencing are expensive, they homogenized subsamples of three to five scats from each cat into a single pooled sample, then compared pooled results against individually extracted samples. The answer was mixed. Common items such as chicken were recovered reliably regardless of pooling, and for two cats the two approaches matched perfectly. But the study’s only wild prey detection—the California vole—appeared only in the individually extracted sample and vanished in the pooled version. When the target of a study is a rare or rarely consumed prey item, the authors conclude, pooling samples before extraction can erase exactly the signal researchers are looking for.

The invertebrate results illustrate another caution. The 16S data were compromised by heavy contamination in negative controls and poor amplification, and although Orthoptera, flies, true bugs, jumping spiders, and noctuid moths were detected in some scats, the team excluded the invertebrate dataset from formal analysis. Primer bias is a further limitation: the 12S primers used here perform poorly on reptiles and amphibians, so the absence of lizard DNA in the scats—despite one owner reporting a caught lizard—may reflect molecular bias rather than true absence. The authors recommend that future studies pair molecular methods with morphological analysis of scat contents to avoid missing entire taxonomic groups.

Despite its small sample size and short duration—cats were sampled for an average of about 12 days—the study demonstrates that owned cats, paired with their willing owners, can serve as a viable study system for predator diet research. Citizen-science sample collection slashes the cost and effort of fieldwork, while owner logs add behavioral context that scat alone cannot provide. And because neither sterilization nor regular feeding reliably suppresses hunting, even well-fed, neutered cats remain relevant to wildlife conservation, particularly where vulnerable prey such as the endangered Amargosa vole, a California vole subspecies, persist. The broader lesson is that a single tool will never fully capture the ecology of a predator that hunts both out of hunger and out of instinct. DNA metabarcoding catches the meals owners never see; owner questionnaires catch the trophies cats bring home uneaten. Used together, they offer the clearest picture yet of what the world’s most popular pet is really doing to the wildlife in its own backyard.

Subject of Research: DNA metabarcoding of owner-collected scat samples to assess the diet and wild prey consumption of owned domestic cats.

Article Title: Using DNA Metabarcoding and Citizen Science to Assess the Diet of Owned Domestic Cats

Article References: Chung, H. J., Vanderzwan, S. L., & Sacks, B. N. (2026). Using DNA Metabarcoding and Citizen Science to Assess the Diet of Owned Domestic Cats. Ecology and Evolution, 16(9), Article e74404. https://doi.org/10.1002/ece3.74404

Image Credits: AI Generated

DOI: 10.1002/ece3.74404

Keywords: domestic cat, DNA metabarcoding, citizen science, diet analysis, wildlife conservation, California vole, pet food, hunting behavior, scat samples, owned cats, Metabarcoding, Citizen