The problem stems from evolutionary history rather than laboratory error. Domestic sheep and European mouflon share a close common ancestry, and the mitochondrial markers most commonly used in carnivore diet studies, such as fragments of the 12S rRNA gene and cytochrome b, are too conserved to separate them. In practice, metabarcoding studies can usually assign Ovis prey only to the genus level, leaving researchers unable to say whether a wolf consumed a protected wild ungulate or a farmer’s livestock. Because these two outcomes carry markedly different ecological and socio-economic weight, misidentification can skew predation assessments, bias estimates of livestock losses and ultimately misdirect wildlife conservation and management efforts, particularly for a recovering large carnivore like the grey wolf (Canis lupus).

A research team working in the French Alps has now developed a solution: a short, diagnostic fragment of the mitochondrial D-loop, a highly variable non-coding control region, that effectively discriminates between the two species. The new marker, named OvisCR, is a 178 base pair amplicon containing 43 single nucleotide polymorphisms. In reference samples, it resolved four mouflon haplotypes and 35 domestic sheep haplotypes with no haplotypes shared between the species. Tested on 115 wolf scat samples in which a general metabarcoding marker had already detected Ovis prey, the fragment achieved species-level assignment for more than 81 percent of scats with Ovis reads, regardless of the bioinformatic pipeline used. The findings, published in Metabarcoding and Metagenomics, provide a robust and practical tool for refining ecological inference in carnivore trophic studies.

Building a reliable diagnostic marker required reference data that public databases simply could not provide. Searches of GenBank revealed that none of the available mouflon D-loop sequences originated from French populations, while none of the numerous sheep sequences with geographic information corresponded to breeds commonly raised in France. No haplotype in the database was exclusive to either species. To close this gap, the team collected 46 European mouflon samples from the Corsican population and four mainland populations in south-western France, alongside 46 domestic sheep tissue samples from the main meat processing plant in the Alps, covering breeds such as Merinos, Lacaune, Mourerous and Préalpes. The mouflon samples combined legally hunted or captured animals with non-invasive faecal and hair collections.

Sequencing the entire D-loop region, roughly 674 base pairs, from these 92 reference samples revealed 45 distinct haplotypes with 82 variable sites and no overlap between species. Haplotype diversity was dramatically higher in domestic sheep, with 40 haplotypes and a diversity index of 0.999, than in European mouflon, with five haplotypes and a diversity of 0.685. This asymmetry likely reflects artificial selection and the diverse geographic origins of sheep breeds, in contrast to founder effects and restricted gene flow in introduced mouflon populations. Rarefaction curves confirmed that all mouflon control region haplotypes in the study area had been detected with a relatively small sample size, whereas additional sheep sampling would likely reveal further haplotypic diversity.

From this alignment, the researchers identified the 178 base pair OvisCR fragment, short enough for metabarcoding of degraded scat DNA yet variable enough to discriminate the species. Most mouflon haplotypes differed from sheep haplotypes by at least three mutation sites, and the flanking regions were sufficiently conserved to design primers with minimal degenerate bases. One exception underscored the evolutionary proximity of the two taxa: a mouflon haplotype, CR178_OM004, differed by only a single mutation from two domestic sheep haplotypes at the short-fragment scale, although the longer 674 base pair fragment separated it by at least six mutations. The team conservatively classified scat samples carrying this haplotype as Ovis sp. rather than assigning a species, prioritising robustness over maximal resolution in a management-sensitive context.

The empirical test used 115 wolf scat samples collected in the same geographic area as the reference samples, all previously found to contain Ovis prey. DNA was extracted and two fragments were amplified in parallel: a 60 to 140 base pair portion of the 12S rRNA gene to confirm Ovis presence, and the new OvisCR amplicon. Sequencing was performed on the Ion Torrent platform with a one-step PCR protocol that reduces the risk of cross-contamination and tag-jumps. A species-specific blocker prevented amplification of wolf DNA in the 12S reaction. The 12S marker confirmed Ovis prey in all 115 samples, and the OvisCR fragment detected Ovis with more than 10 reads in 95.7 to 96.5 percent of samples depending on the pipeline.

A central methodological contribution was the comparison of two widely used bioinformatic pipelines: OBITools, which clusters sequences into operational taxonomic units using dissimilarity thresholds, and DADA2, which infers amplicon sequence variants through explicit error modelling. Both pipelines demonstrated similar accuracy, with the OvisCR fragment achieving species-level assignment for more than 81 percent of scats with Ovis reads detected. DADA2 consistently yielded slightly more species-level assignments, particularly at lower read thresholds, likely due to its denoising strategy, while the OTU-based OBITools approach proved more conservative for samples with limited read depth. Discrepancies between pipelines were rare and primarily concerned cases with few OvisCR reads rather than conflicting species assignments, highlighting the robustness of the marker across analytical frameworks.

Read-count thresholds emerged as a critical, yet often arbitrary, step in interpretation. Negative and positive controls showed that contamination levels for the OvisCR fragment were consistently low, never exceeding 10 Ovis reads, which justified exploring a lower threshold of 10 reads per sequence per sample compared with the conservative 100-read threshold applied to the universal 12S marker. Lowering the threshold increased the number of species-level assignments without introducing clear mis-assignments. Samples assigned only at the genus level had significantly fewer reads than those assigned to species, suggesting that taxonomic resolution was partly constrained by sequencing depth rather than by the marker itself. The impact of thresholds depended more strongly on the molecular marker than on the choice of pipeline, a pattern that contrasts with parts of the metabarcoding literature and warrants further investigation.

From a conservation and management perspective, the ability to discriminate between predation on domestic sheep and European mouflon is essential. Domestic sheep predation is a central driver of human-wolf conflict in France and elsewhere, while predation on mouflon, though relevant for game management, carries different implications. Misclassifying one species as the other could lead to incorrect estimates of livestock depredation, misinformed conflict mitigation strategies and flawed compensation or culling decisions. The OvisCR approach can serve as a model for similar developments targeting other taxonomically challenging prey groups, though the authors caution that the diagnostic value of the fragment should not be assumed transferable beyond the French context. Broader geographic applicability requires prior validation using region-specific reference datasets encompassing the diversity of domestic sheep, European mouflon and other Ovis species, a reminder that in metabarcoding, local reference data remain the foundation of every reliable species assignment.

Subject of Research: Genetics

Article Title: A 178 bp diagnostic D-loop fragment (OvisCR) to discriminate domestic sheep ( Ovis aries ) and European mouflon ( Ovis musimon ) in diet metabarcoding studies

Article References: Pirog, A., Lavarec, J., Rousselot, S., Kaerle, C., Duchamp, C., & Queney, G. (2026). A 178 bp diagnostic D-loop fragment (OvisCR) to discriminate domestic sheep (Ovis aries) and European mouflon (Ovis musimon) in diet metabarcoding studies. Metabarcoding and Metagenomics, 10, Article e172476. https://doi.org/10.3897/mbmg.10.172476

Image Credits: AI Generated

DOI: 10.3897/mbmg.10.172476

Keywords: metabarcoding, grey wolf, DNA diet analysis, mouflon, sheep, mitochondrial D-loop, conservation, diagnostic, D-loop, fragment, OvisCR, discriminate

The research, led by Kat Douglas of Georg-August-Universität Göttingen together with Zahra Ebrahimi Monfared, Seyedhamed Mirkarimi, Adrian Paterson, and Mahmood Soofi, drew on face-to-face interviews with 292 residents across 30 villages lying within five kilometers of the park boundary. Interviews were conducted between April and November 2022 by two local field researchers, one male and one female, with 71 percent conducted in Turkmen and the remainder in Farsi. Respondents included livestock herders, crop farmers, beekeepers, teachers, and freelancers, reflecting the mixed livelihoods of a landscape where Turks, Kurds, Fars, Baluch, and Sistani communities live alongside some of Iran’s most threatened mammals.

Golestan National Park is a fitting laboratory for such questions. Established in 1957 as Iran’s first biosphere reserve, it covers roughly 91,895 hectares and rises from 450 to 2,411 meters, encompassing Hyrcanian temperate forest, steppe, and semi-arid habitats. The park shelters urial, bezoar goats, red deer, and roe deer, and with them the leopards, wolves, and bears that prey on them or, in the bear’s case, raid the crops and hives of surrounding farms. As human populations and agriculture have expanded through the Hyrcanian forests, negative interactions have intensified, and retaliatory or precautionary killing of carnivores remains a common response to livestock loss.

What sets this study apart is its analytical framework. Rather than treating attitudes as a simple product of economic damage, the team used Bayesian structural equation modelling, a statistical technique capable of estimating multiple direct and indirect pathways simultaneously among latent constructs such as beliefs, norms, and attitudes that cannot be measured directly. The workflow proceeded in two steps: first a confirmatory factor analysis verified that survey items reliably measured their intended constructs, with all retained items loading above 0.40 and reliability coefficients such as McDonald’s omega ranging from 0.78 to 0.83; then a structural model tested how cognitive beliefs, social beliefs, perceived management responsibility, and livestock loss shaped attitudes toward each species separately.

The headline finding is strikingly consistent. Across all three species, cognitive beliefs, the individual’s evaluations, knowledge, and perceptions of carnivores, emerged as the strongest and most reliable predictor of attitudes. For bears the standardized path coefficient reached negative 0.72, for wolves negative 0.63, and for leopards negative 0.48, with 95 percent Bayesian credible intervals that excluded zero in every case. Respondents who felt carnivore conservation was unimportant, or who believed that any carnivore attacking livestock should be killed, held markedly more negative attitudes overall. In other words, the mind, not the wallet, appears to be the primary battleground of coexistence.

Social beliefs, the expectations of family and fellow villagers about retaliatory killing, also pushed attitudes in a negative direction, though their direct effect was weaker and, in the leopard model, the credible interval overlapped zero. More intriguingly, social beliefs were positively associated with cognitive beliefs in all three models, suggesting that when peers endorse killing carnivores, individuals’ own perceptions of carnivores as dangerous are amplified. The authors caution that social desirability bias, in which respondents answer as they believe their community expects, may partly shape such patterns, a well-known hazard of attitudinal surveys. Even so, the indirect pathway points to a clear lever: shifting community-level norms could ripple through individual belief systems and soften attitudes toward predators.

Perhaps the most counterintuitive result concerns livestock loss itself. Included as an observed variable in the leopard and wolf models, it showed only a weak, statistically uncertain relationship with attitudes, with coefficients near zero and credible intervals spanning zero. Reducing the material costs of coexistence, while still vital for household welfare, appears insufficient on its own to change how people feel about carnivores. The bear model dropped livestock loss entirely, a logical choice given that brown bears in northern Iran mainly raid orchards and beehives; one study found 44 percent of negative interactions involved orchards and 32 percent beehives, against just 13 percent livestock attacks. Notably, human-caused mortality records from Iran between 1980 and 2021 show bears were killed more often than wolves or leopards, but chiefly over threats to human life and crops rather than depredation of herds.

Perceived management responsibility added another layer of nuance. The belief that government should capture and relocate carnivores after an attack was associated with more positive attitudes, an effect that was statistically credible for leopards and wolves, though modest in magnitude. Yet in the wolf and bear models this same variable correlated with more negative cognitive beliefs, hinting that institutional expectations and personal belief systems operate somewhat independently. The context matters: in Iran, compensation schemes for wildlife damage are widely described as bureaucratic and rare, and the survey confirmed this, with only 1.7 percent of respondents, five people, reporting ever receiving compensation. Post-attack support limited by funding and paperwork is known to contribute to illegal retaliation.

Several hypothesized drivers refused to stay in the models. Emotions such as fear, pride, and happiness were removed after destabilizing the analysis, though exploratory estimates hinted that positive emotions favored positive attitudes. The authors suspect social desirability bias may be at play, since roughly half of respondents reported minimal fear of leopards, wolves, or bears, a pattern that could reflect genuine tolerance or reluctance to appear afraid. Demographics including age, gender, ethnicity, literacy, and occupation were also excluded, as were legal and participatory variables, the latter perhaps reflecting the fragmented legal framework surrounding conservation in Iran. Women were underrepresented at 16 percent of the sample, a limitation rooted in cultural barriers rather than sampling design, and the authors acknowledge that the final models were deliberately simple, with two items per construct.

The practical implications reach well beyond Golestan. Because cognitive beliefs dominate attitude formation, education programs that explain the ecological role of carnivores, recalibrate perceived danger toward realistic risk, and teach practical protection, from electric fencing and guardian dogs to fox lights and other non-lethal deterrents, may do more for coexistence than compensation checks alone. Because social norms feed individual beliefs, community-wide engagement rather than one-on-one persuasion is likely to be decisive. And because trust in management shapes attitudes, transparent, participatory governance deserves a place alongside any mitigation toolkit. The authors emphasize that their findings are local in origin but plausibly generalizable in mechanism, since belief systems, social norms, and institutional perceptions recur across human-carnivore contexts worldwide. Coexistence, this study suggests, is negotiated less in pastures and orchards than in the shared beliefs of the people who live there.

Subject of Research: Local community attitudes toward large carnivores around Golestan National Park, Iran

Article Title: Claws, consequences and coexistence: Local community attitudes toward large carnivores in mountain landscapes

Article References: Douglas, K., Monfared, Z. E., Mirkarimi, S., Paterson, A., & Soofi, M. (2026). Claws, consequences and coexistence: Local community attitudes toward large carnivores in mountain landscapes. Ambio. https://doi.org/10.1007/s13280-026-02486-9

Image Credits: AI Generated

DOI: 10.1007/s13280-026-02486-9

Keywords: large carnivores, human-wildlife coexistence, Persian leopard, grey wolf, brown bear, Golestan National Park, conservation psychology, Bayesian structural equation modelling, social norms, cognitive beliefs, livestock depredation, Iran