Ecology

AbstractLow wheat productivity in Ethiopia is largely attributed to nutrient imbalances and the widespread use of blanket fertilizer recommendations that do not account for site-specific soil conditions. To identify the most yield-limiting nutrients for wheat (Triticum aestivum L.) production, a field experiment using nutrient omission trials was conducted during the 2021 cropping season at eight farmers’ fields in Wera Dijo District, Halaba Zone, central Ethiopia. The soils across the study sites had neutral pH (6.1–7.0), medium to high organic carbon content (1.85–4.19%), and low to medium available phosphorus (3.10–10.94 mg kg⁻1). The experiment was arranged in a randomized complete block design with ten treatments, including a complete nutrient treatment, NP treatment, negative control, and single-nutrient omission treatments. Grain yield, biomass yield, and yield components were analyzed using a linear mixed-effects model, with nutrient treatments as fixed effects and farms and blocks within farms as random effects. Nutrient treatments significantly affected grain yield, biomass, and yield components (p < 0.0001). Omission of nitrogen or phosphorus caused significant reductions in all measured parameters, with nitrogen omission resulting in the greatest yield loss, comparable to the negative control, indicating that nitrogen was the most limiting nutrient. In contrast, omission of potassium, sulfur, zinc, or boron had no significant effect on wheat yield, suggesting that these nutrients are not currently limiting in the study area. Grain yields under the NP treatment were statistically comparable to those under the complete treatment. These results highlight nitrogen and phosphorus as the primary yield-limiting nutrients for wheat production in the study area and demonstrate the importance of site-specific nutrient management. Targeted application of N and P fertilizers can improve wheat productivity and nutrient use efficiency in Halaba and similar agro-ecological zones.

AcknowledgementsAcknowledgements The authors acknowledge the Ethiopian Institute of Agricultural Research (EIAR) for funding this study. We are also grateful to the experimental farmers for providing land and supporting the implementation of the field experiments.FundingThis research was funded by the Ethiopian Institute of Agricultural Research (EIAR).Author informationAuthors and AffiliationsHawassa Agricultural Research Center, Sidama Agricultural research Institute, Hawassa, EthiopiaAbay Ayalew & Moges TadeseAuthorsAbay AyalewView author publicationsSearch author on:PubMed Google ScholarMoges TadeseView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
Abay Ayalew.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleAyalew, A., Tadese, M. Identifying limiting nutrients for wheat (Triticum aestivum L.) production in Halaba, Central Ethiopia.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-51398-xDownload citationReceived: 02 November 2025Accepted: 27 April 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51398-xShare this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsNutrient omissionMacro- and micronutrientsNutrient recommendationYieldPenalty More

AbstractAnimal personality influences organismal interactions and individual habitat use. Rodents are zoonoses reservoirs and often exposed to several pathogens simultaneously, potentially resulting in interdependence of infections and susceptibility to infection. Still, entire pathogen communities are rarely investigated, even though, given rodents ubiquity in human settlements, understanding the link between animal personality and pathogenesis is an important public health issue. We investigated the association of animal personality with pathogen communities in wild rodents, analysing ectoparasite occurrence and pathogenic bacteria of 93 individuals belonging to 3 species from urban and forest areas around Potsdam, Germany. Individual personality was quantified using a combination of open-field and dark-light test. Rodents were then euthanised and screened for pathogens in the spleen through 16 S rRNA amplicon sequencing, and ectoparasites were collected. We detected 6 pathogenic bacteria and 3 ectoparasite taxa. Host species and sampling time explained most of the variation in pathogen associations, but within each genus, 7–9% of the variation was explained by animal personality. Active rodents were more likely infected by Bartonella than less active ones. Bold animals had lower tick infestation probabilities. Thus, animal personality contributes to the distribution and prevalence of pathogens in wild rodents, and should be considered in epidemiology and disease management.

IntroductionZoonotic diseases are an emerging global threat1and understanding transmission patterns and pathogen occurence and prevalence within the animal population is critical. Variation in pathogen prevalence among individual animals may not only depend on the individual immunosystem, but also on a range of behavioural correlates that may affect space use and sociability of animals. For example, parasite load and parasite transmission can relate to behavioural traits (e.g2,3.,). More exploratory chipmunks (Tamias minimus) host a greater abundance of ectoparasites4. Bolder grey squirrels (Sciurus carolinensis) are more likely to be infected by gastro-intestinal helminths5. Less explorative multimammate mice were more likely to be infected by MORV virus6 in relation to their more exploratory conspecifics. Conversely, the presence of certain pathogens can influence animal behaviour. Toxoplasma gondii, for example, alters behaviour in mice making them bolder7; honeybees (Apis mellifera) alter their behavioural physiology after being infected by Nosema ceranae8, and exposure to the bacterium Serratia marcescens during development affects the expression of boldness in crickets9. These examples highlight the strong links and feedbacks between host behaviour and pathogen prevalence.The concept of animal personality may capture some of the behavioural variation linked to pathogen prevalence. Animal personality is commonly defined as the between-individual differences in behaviour that persist through time and contexts10,11. Animal personality can also be linked to fitness and can be partly heritable10. Personality traits are particular aspects of an individual’s behavioural repertoire12, and out of these traits, five major categories including boldness, aggressiveness, activity, exploration, and sociability have been studied extensively13,14. However, not much is known about how animal personality is linked to the composition of pathogen communities, and to which single pathogens, even though this could crucially improve understanding disease transmissions and their spread in populations. Especially when considering that only a small proportion of animals is responsible for a large proportion of the transmission of a pathogen (“80/20 rule”15) due to the typical negative-binomial distribution of infection severity and parasiste loads, it might be crucial to understand which traits distinguish individuals that are highly infected. Personality traits such as sociability, exploration and boldness have been associated to this asymmetry of transmission and infection. These traits have been focused on independently showing their separate effects on diseases transmission, as well as considered together as suits of correlated behavioural traits16. When focusing on the single effect of exploration, an experiment on little brown bats (Myotis lucifugus) using a fluorescent powder as a substitute pathogen showed that the more explorative male bats were more likely to transmit and acquire infections3 than the less explorative males. In domestic cats boldness was positively linked to the infection probability of Feline Immunoficiency Virus, a lethal disease17. Furthermore, personality traits can alter where animals shed infectious agents in the environment15 as space use is often linked to animal personality18 and can shape whether individuals spatially interact with those of other species’19. Sociability may affect contact rates, which are important for microbiome diversity20 but may also affect pathogen exchange, direct disease transmission and infection patterns, as shown in three-spined sticklebacks21.In deer mice (Peromyscus maniculatus), bolder individuals were more often in contact with conspecifics and thus also more likely to be infected by hantavirus22 and in Eastern grey squirrels bolder and more explorative individuals were more likely infected by gastro-intestinal helmiths5. These examples show that animal personality traits affect pathogen loads, encounter rates and patterns of disease transmissions and that not only effects of single personality traits are relevant but that suits of correlated behavioural traits, i.e. behavioural syndromes, play an important role as well, as several traits might act together in creating the observed infection patterns. Here we aim to study an entire pathogen community, and relate occurrence of the single pathogens as well as community richness and composition to animal personality in forest rodent communities.Ectoparasites are known vectors for zoonotic infectious agents23,24. Links between animal personality and ectoparasite presence are equivocal: bolder ground squirrels carried more ticks4, but shyer bank voles tended to use higher vegetation cover and carried a higher tick load than bold conspecifics25. More explorative firebugs had higher loads of ectoparasitic mites26. Ectoparasite load alone can impact fitness and overall health27,28, linking them to pathogen transmission for their hosts and humans29. Animal personality may in part explain the likelihood of acquiring ectoparasites. In most small mammal species investigated so far, bolder or more exploratory individuals had larger home ranges30 or larger ranges but less overlap with conspecifics in these ranges18,31 than shyer or less explorative individuals. Considering that most ectoparasites are free-living, personality-mediated space use and microhabitat choice may contribute to higher ectoparasite loads through an increase in host-ectoparasite encounters. We therefore added ectoparasite infestations to the analysis of pathogen communities in forest rodents.Rodents are the most abundant and diversified group among mammals, representing more than 40% of mammalian species32. They are reservoir hosts for multiple zoonotic diseases, of which to date 66 were identified33. Their worldwide distribution and proximity to human settlements make the risk of zoonotic spillovers to humans highly likely. Rodent ectoparasites have been identified as hosts for agents responsible for diseases such as ehrlichiosis, Lyme disease, tick-borne encephalitis, bubonic plague, and murine typhus among others23,24. To understand prevalence and pathogen dynamics in this group, the co-variation of pathogens and animal personalities may thus be important for disease prevention.In this study we take a holistic approach trying to understand how host behaviour affects infection risk in natural zoonotic reservoirs. This not only would improve our theoretical understanding of disease dynamics but also would contribute to advance epidemiological surveillance and disease management. To achieve this, we applied molecular methods to detect multiple pathogens with no a priori knowledge, from free living rodents in parks and gardens. We also collected and counted ectoparasites, which can be involved in transmission of bacterial infectious agents. To add knowledge on behavioural drivers of pathogen occurrence, we collected individual-based, behavioural information with standard behavioural tests and repeated testing. We hence quantified consistent individual traits, that might collectively modulate infection risks across multiple pathogens simultaneously. We made use of an ongoing rodent collection a pan-european surveillance study investigating rodent pathogen diversity in parks and forests across Europe (BiodivERsA-BioRodDis, https://www6.inrae.fr/biodiversa-bioroddis) and added an additional, in-situ behavioural information layer for each individual, before the terminal collection of the rodent. We hypothesized that single pathogen infections and pathogen communities vary among animals with different behavioural phenotypes, and that the richness of pathogen communities may correlate to personality axes. We expect bolder and more active individuals of all species in the rodent community to have the higher infection probability for pathogens and ectoparasites, compared to shyer and less active individuals, due to their larger home ranges increasing the chance for infection, and higher contact rates with conspecifics.ResultsBehaviourIn total 186 individuals, 79 Clethrionomys glareolus, 56 Apodemus agrarius and 51 A. flavicollis, were captured in a park and a forest near Potsdam, Germany, and their behaviour was quantified. Both activity and boldness differed between the two genera, with Apodemus being bolder (more likely to emerge from shelter) and tended to be more active (Table 1, A4, A5 and Fig. 1) than Clethrionomys glareolus. Behavioural variables did not differ between the two Apodemus species (activity: Chi2 = 0.528, df = 1, p = 0.468; boldness Chi2 = 0.01, df = 1, p = 0.918; Table A5).Table 1 Behavioural variables from an emergence test (emergence from shelter in 5 min, representing boldness) and an open field test (proportion of floor sections covered in 5 min, representing activity) in 186 forest rodents analysed with generalized linear mixed models. Reference for season was autumn, for site was park, for genus was apodemus, for sex was female. Repeats were numbered (first to third test, however not all individuals were tested repeatedly). Individual ID was used as random factor.Full size tableFig. 1The alternative text for this image may have been generated using AI.Full size imageBehavioural variables chosen as quantitative measures for boldness (latency head) and activity (proportion of floor sections explored) in 93 wild forest rodents in their first test. Dots represent the values of individuals (scatterplot), boxplots and dot colours the species Clethrionomys glareolus (Cle, 41 individuals), and the genus Apodemus (Apo, 47 individuals).Pathogen and ectoparasitesWe detected pathogens in the spleen of 71 animals, and both spleen pathogens and ectoparasites in 86 out of 93 animals. In 22 animals no pathogens were detected, 83% of these were C. glareolus. Uninfected C. glareolus were found at both study sites while uninfected A. agrarius were only found at the park site, and uninfected A. flavicollis were only found at the forest site (Table 2). Pathogens correspond to 6 bacterial genera and one Apicomplexa family (Sarcocystidae). Mycoplasma haemomuris (mych), Bartonella (bart), and Candidatus Neoehrlichia mikurensis (neom), were commonly detected in both rodent taxa, while Mycoplasma coccoides (mycc) was detected only in the genus Apodemus, while Sarcosystidae (sarc) was found mainly in C. glareolus. Francisella Orientia, and Anaplasma were negative in all tested individuals. Table 2 Ectoparasite and pathogen occurrence in percent of samples from three species of forest rodents (Aagr: Apodemus agrarius; Afla: Apodemus flavicollis; Cgla: Clethrionomys glareolus) in Germany sampled in autumn 2021 and spring 2022 in a park and a forest habitat. D, darker cells indicate higher occurrence.Full size tablePathogens and behaviourThe composition of the community of pathogens in the rodent spleen were affected by behavioural types, but none of the behavioural variables did explain the richness of the pathogen community (Table 3). Richness of pathogens and ectoparasites combined was higher in Apodemus mice, compared to Clethrionomys voles, higher in males compared to females, and higher in the forest site compared to the park site (Table 3). Between forest rodent species, richness of the pathogens and ectoparasites combined was higher in A. flavicollis compared to C. glareolus (estimate = 0.41, p = 0.021), but not compared to A. agrarius (estimate = −0.10, p = 0.842). C. glareolus and A. agrarius did not differ (estimate = 0.32, p = 0.157) (Table A7).Table 3 Pathogen richness with and without ectoparasites considered (A), pathogen community composition (B) in the spleen of forest rodents, with regard to rodent genus (Apodemus vs. Clethrionomys), sampling time (spring vs. autumn) and sampling site (forest vs. park site), sex (male vs. female) and animal behaviour, with pro. sections (proportion of floor sections covered in an open field test) coding for activity, and emergence from shelter in a dark-light test coding for boldness.Full size tableThe composition of the community of pathogens in the rodent spleen and ectoparasites (based on their occurrence data) differed among rodent genera (Permutational Multivariate Anova, F = 6.3, p = 0.001, Table 3B, Fig. 2E) explaining 11% of the variation, between the sampling times (F = 7.2, p = 0.002, 8% explained, Fig. 2D), and was related to the activity of the individual (F = 3.0, p = 0.046, explaining 3% of the variation). Within the pathogen community of each separate genus, activity explained 9% (Apodemus), or 7% (Clethrionomys), respectively (Fig. 2F, Table 3B). Boldness did not explain the pathogen community in either genus. In Apodemus, sex additionally explained 10% of the variance in the pathogen community composition. For C. glareolus, 20% of the variance in the pathogen community was explained by sampling season (F = 8.8, p = 0.001).Fig. 2The alternative text for this image may have been generated using AI.Full size imageOrdination of combined pathogen and ectoparasite communities in 71 rodents from a forest and a park site (non-metric dimensional scaling plots (Stress = 0.054, clusters k = 4) (A) Pathogens found in the spleen (Mycoplasma haemomuris (mych), M. cocoides (mycc), Bartonella (bart), Candidatus Neoehrlichia (neom), Sarcocystidae (sarc), Borrelia (borr)) and ectoparasites found on the animal (fleas, lice and ticks). (B) Pathogen and Ectoparasite communities plotted together with rodent individuals (numbers), C-F) grouping of individuals by (C) sampling sites (D) sampling season, (E) host genus (Cle: Clethrionomys; Apo: Apodemus) and (F) by activity type.More active animals in both rodent genera were more likely to be infected with the most common pathogen, Bartonella sp. (33–65% of individuals per species infected, Table 2). Sarcocystidae, which occurred only in C. glareolus, were also positively related to activity, i.e., more active voles were also more likely to be infected (Table 4). Other pathogens were not affected by the activity of the individual (Table 4). The variable representing boldness did not have any effect on pathogen occurrence (Table 4).Table 4 Overview of factors affecting single pathogen occurrence in the spleen of 93 forest rodents, with regard to rodent genus (Apodemus vs. Clethrionomys), sampling season (spring vs. autumn) and sampling site (site 1 (a forest) vs. site 2 (a park)), sex (female vs. male) and individual behavioural type (proportions of sections covered in an open field test is taken as a measure of activity, emergence from shelter in a dark light test was used as a measure of boldness).Full size table The probability for a rodent to be infested by ectoparasites was influenced by individual behaviour, and differed among species and genera (Table 4 and A6, A7). More A. agrarius individuals had fleas compared to C. glareolus (effect estimate = 2.19, p = 0.019) and tended to have also more fleas compared to A. flavicollis (estimate = 2.04, p = 0.058) individuals. C. glareolus and A. flavicollis did not differ regarding their flea occurrence (estimate = 0.15, p = 0.971) but regarding their tick occurrence, with more A. flavicollis having ticks compared to C. glareolus (estimate = 3.75, p = 0.003; Table A7). Tick occurrence did not differ between the two Apodemus species (estimate = −2.15, p = 0.172) but tended to differ between A.agrarius and C. glareolus (estimate = 1.60, p = 0.091; Table A7). For all rodents species, tick occurrence was positively correlated to emergence, i.e., the longer it took until the animals’ head appeared in the opening (the longer = the shyer), the more likely it was that ticks were found (Table 4).Occurrence of many ectoparasites and pathogens (i.e., fleas and ticks, Candidatus Neoehrlichia mikurensis and Borrelia sp.) was higher on the forest site, compared to the park (Table 4, Table A6). Differences between males and females were observed in the occurrence of Mycoplasma haemomuris, and Bartonella sp., with males having higher occurrences of the pathogens than females in both cases (Table 4). Sarcocystidae and Borrelia were rare in Apodemus (2–12%, depending on species). In C. glareolus, Sarcocystidae was found only during spring (10 cases; Estimate = 2.62, p = 0.015), and in Apodemus, Borrelia were only found at the forest site (5 cases, Table 2; Estimate = 2.24, p = 0.046; Table A3).DiscussionUnderstanding the impact of host behaviour on infection patterns in wild reservoirs hosts is an important step towards understanding disease dynamics, which can have direct benefits for public health by improving epidemiological surveillance and disease management strategies. Thereby it is important to consider that individual behaviours of hosts may interact to influence the infection risk for multiple pathogens simultaneously, rather than affecting each pathogen individually. Here we applied this integrative framework by quantifying two different individual behavioural traits, and investigating how these affect pathogens found in the spleen, as well as ectoparasites of two rodent genera. We found that boldness only influenced the probability of individuals being infested by ticks, while individual activity affected community composition of pathogens as well as the probability of individuals being infected with Sarcocystidae and Bartonella. Although most variation in infection patterns was explained by host genus, season or sampling site we could still show that animal personality contributes to the distribution and prevalence of pathogens in wild rodents.Pathogens found in this study are typical for rodents inhabiting forests in Europe: Bartonella can be pathogenic on both rodents and humans34,35–36, Mycoplasma is a common parasite of mammals37 including Apodemus mice and C.glarolus3839. M. haemomuris can cause infectious anaemia in rodents40. We did not find support for associations of Mycoplasma to behavioural traits. Mycoplasmae are among the smallest self-replicating prokaryotes37 and can spread airborne for many kilometres41.Candidatus Neoehrlichia sp., was present in ~ 25% of all individuals in our study, matching earlier findings42,43–44. Although related Neoehrlichia mikurensis causes inflammatory disease in humans, infected rodents seem asymptomatic44. Neoehrlichia sp. can be transmitted by ticks45, however the probability to have ectoparasites was not related to the infection in our data. Further, Borrelia was not related to behaviour, but we have to be careful with this genus since the spleen is not the target organ to detect Borrelia infections.We found that Bartonella had higher levels of prevalence in more active individuals, and in males. Bartonella seems to be abundant in rodents and their fleas46. High activity levels may increase encounters with conspecifics and their fleas, and thus put individuals at a higher risk of contracting Bartonella.This finding is in accordance with sex differences we detected in both Bartonella and Mycoplasma, with a higher occurrence in males than in females. Males in most mammals have larger home ranges and more interactions with conspecifics than females, and often also carry higher flea loads (e.g47), which increases the risk of Bartonella contraction. Meanwhile, in a study on Microtus voles, Bartonella infections were not linked to direct contact rates20, indicating that individual activity may translate to other behaviours, e.g. space use or the use of joint nests or points of interest, where fleas can be contracted. However, while the probability to be inflected with Bartonella was dependent on individual activity, the probability to be infested with fleas was not influenced by activity in our study. This suggests that there is no influence of animal personality on the likelihood of encountering fleas and that the latter is also independent of exposure to Bartonella. Thus, the positive association between Bartonella and individual activity might not reflect differences in encounter rates with fleas, but rather individual variation in host susceptibility. This would align with the findings of Koprivnikar et al.48 who could show an association between a behavioural syndrome, formed by activity, boldness and exploration, and the susceptibility to parasite infection in tadpoles.Ectoparasite infestation can positively correlate with personality traits, for example with boldness in ground squirrels45, and exploration in great tits49. However, encounter probability with ticks and other ectoparasites might not always be directly linked to personality traits but could be indirectly resulting from personality-dependent social interactions, or space use. For example, bolder C. glareolus and A. agrarius had larger home ranges with higher levels of ground cover (i.e., vegetation cover in 10 cm from the ground) compared to shyer conspecifics1819. These personality-dependent microhabitat preferences could in turn affect encounter rates with ticks, as the tick larva, which preferably infest rodents, occur primarily in lower vegetation layers (0–9 cm50). In great tits (Parus major) more explorative individuals had more social associations compared to their less explorative counterparts51 which could influence parasite encounter rates. Higher numbers of contacts to conspecifics might translate to higher chances of encountering with parasites infested individuals or areas where parasites occur, as it was shown for sleepy lizards (Tiliqua rugosa) where bolder individuals had more social interactions and a higher probability to be infested with ticks52.In our study shyer rodent individuals surprisingly had a higher probability to be infested by ticks compared to bolder individuals, which was also found in an earlier study on C. glareolus in Sweden (Erixon et al. submitted). This might be correlated to the finding that shyer rodents use areas of higher maximum vegetation height, i.e., areas characterized by bushes and trees18. Such areas might be more attractive habitats for some local ticks, such as Ixodes Ricinus, and thus, might represent areas with higher tick abundances53. The overlap of areas preferred by shy individuals with areas of higher tick abundances might result in higher chances of encounters between hosts and parasite and might explain the observed pattern of probability of tick infestation between bold and shy rodents. Rather than just vegetation density, vegetation types might be also relevant for the encounter probability between ticks and hosts species in a certain area, as well as species-specific habitat preferences of the parasites. Since ticks were not determined to species in our study, we cannot say anything regarding species specific effects, but it seems likely that different tick species might occur in some habitat areas in higher numbers compared to other areas. The specific relationship between boldness and tick infestation probability might thus be a result of species-specific non-random distribution patterns of ticks paired with non-random, personality driven distribution patterns of the host species. Depending on the species observed the overlap of these patterns might change explaining why contradictory findings for the effect of boldness on tick infestation probability exist.Animal personality traits have been shown to correlate with individual space use54, spatial overlap with conspecifics and heterospecifics (e.g18,19), and fitness proxies (e.g13) and can significantly affect the likelihood of an individual contracting infections. It is known that a small proportion of individuals contribute disproportionately to the transmission of a pathogen15 nowadays called “superspreader”. Both sociability and boldness have been associated to this asymmetry of transmission, as well as activity and exploration. Infection patterns have often been reported to be biased towards males55 but when considering personality and sex together patterns are less clear. For little brown bats it was for example shown that the bias in parasite infection towards one sex changed depending on the parasite species. The positive effect of exploration on the infection probability on the other hand was the same in both sexes56. Thus, the asymmetry of infection related to the interaction of sex and personality traits might be highly pathogen/parasite specific. Parasitized individuals may differ in their approach to predators (e.g57,58,59.) or the general level of activity60,61–62 but it is often unclear whether the differences are due to behavioural changes induced by the pathogen, or were the pre-condition to contract pathogens. The rodents in our study were more likely to carry Bartonella spp if they were more active. Effects of activity were picked up for pathogen composition (Table 3A), however the community effect may have been largely driven by the occurrence of Bartonella in both genera, and the occurrence of Sactocystes in voles.ConclusionHere, we demonstrate that animal personality traits can contribute to the distribution, prevalence, and co-occurrence of pathogens in wild rodents. Active individuals carried more pathogen species and a different pathogen community then less active individuals. Hosting multiple pathogens simultaneously may impose increased immune demands, potentially resulting in trade-offs or immune suppression that further influence susceptibility to additional infections. These findings highlight the importance of incorporating individual behavioural variation into epidemiological frameworks and disease management strategies.MethodsRodents, sites and capture protocolRodents were sampled in October 2021 and May 2022. We collected both behavioural and pathogen data from 99 individuals, belonging to the species bank vole Clethrionomys glareolus (46 individuals), yellow necked mouse Apodemus flavicollis (26), striped field mouse Apodemus agrarius (21), and common vole Microtus arvalis (6). The latter was excluded from the analyses presented here (n = 93, two genera) due to small sample size. Sampling was replicated at two sites, a botanical garden in a semi-forested, urban park (290 ha) in Potsdam, Brandenburg, northeast Germany, surrounded by urban settlements with a mixture of sealed and wooded areas and a constant human presence (park site); and a forested area at the edge of a large forest (875 ha), 3 km distant from the first site and composed of mixed coniferous forests, meadows, train tracks and a main road, but no walking paths (forest site). Details of the area and rodent capture methods are described in Firozpoor et al.63. Trapping was conducted on both sites in each sampling time. In short, we captured, marked and released rodents for a week conducting behavioural tests, and the next week we collected and sacrificed these individuals for the pathogen sampling. In total we conducted 164 animal personality tests (206 in total, of which 19% were second and 4% third tests). Some individuals were tested but not re-captured when collecting the individuals for pathogen sampling, but we included their behavioural data to describe the variability and repeatability of the behavioural data collection.Rodents were captured using a combination of live traps (Longworth and Ugglan traps) with a 1 cm hole to allow shrews to escape64 since these were not target species. At each site a total of 100 inactive traps were placed in 4 transects with 25 m spacing among transects and 10 m between traps and, pre-baited for 2–3 days. After pre-baiting, traps were equipped with fresh bait and nesting material and activated in the evening. The following morning empty traps were deactivated. Captured animals were kept in their traps and provided with fresh food and nesting materials. Lactating and pregnant females were not included in the testing and were directly released on the capture location. The other captured animals were transported in their trap up to 100 m from their point of capture to the behavioural testing arena at a shady location on the site and tested during the morning. After the test, animals were weighed and individually fur-marked using a small scissor, and subsequently released at the location of capture during the first week, or euthanised during the second week. The proportion of animals tested repeatedly varied among the species (25% of C. glareolus, 39% of A.agrarius, and 29% of A. flavicollis had repeated tests).Behavioural testingWe combined two standard tests, the dark-light and the open-field test18. Both tests were video recorded from above using Cam Park Action Cameras X15 and scored afterwards. The dark-light test measures the willingness of an individual to leave a dark protected shelter to enter an open, well-lit and potentially dangerous new space65. The setup consisted of an opaque PVC tube (length 32 cm, diameter 10.5 cm, closed by a slightly smaller cylinder) connected to a plastic arena that emulated an exposed area (Figure A1 A and B). The entrance leading to the arena had an outer opaque door and an inner, transparent door. The latter was designed as a one-way, self-closing door only opening towards the arena and closing once the individual passed it, thereby preventing re-entering of the tube (Figure A1 C). Rodents were first brought inside the PVC tube and left to acclimatize in the dark for a minute. Afterwards the outer door leading to the arena was opened, letting light in, and a 300 s (s) timer started. The animal now could either stay in the tube or push the inner door and enter the arena. Two latencies were recorded during this test: latency head describes the elapsed time until the animal first peaked its head until the neck into the arena. The Latency body describes the time until the animal enters the arena with its full body excluding the tail and concludes the dark-light test. If the rodent did not leave the tube, the test concluded at the 300 s mark, and the rodent was assigned a latency of 300 s for both measurements. In this case, the rodent was carefully displaced by the operator into the arena, either by slowly rotating the entrance tube (first step), or by gradually reducing the space in the tube by slowly pushing the smaller cylinder into the tube (second step).As soon as the animal had entered the arena with its full body, the open-field test started. This test measures the exploratory behaviour and the general activity of an animal within a delimited open space6667. We used a circular arena with a diameter of 120 cm surrounded by 60 cm high walls and a net on top. Drawings on the arena bottom divided the arena into an inner (45 cm wide) and outer (15 cm wide) section, and 8 sectors, meeting running through the center, further divided the arena into 16 sections. The test ran for 300 s. During this time, five measurements were recorded: (1) Latency center, which describes the time in seconds until the animal first steps into the inner circle on its own initiative, with its full body excluding the tail, and it defaults to 300 s if this condition is not met (cases where the animals fled the dark-light entrance tube when we tried to displace it carefully, and thereby crossed the line, were not counted). (2) The proportion of visited sections with the full body excluding the tail. (3) The frequency of jumps (total number of jumps/5 min) (4) The crossing frequency (total number of crossings into the inner ring/5 minutes). (5) Every ten seconds activity was recorded as a binary value, wherein the animal was currently active or inactive. A rodent was defined as active if any type of locomotion was displayed.Pathogen and ectoparasite detection The week following personality tests, traps were activated at the same locations every evening, and checked daily for a week. Captured rodents were transported inside of their traps to the Animal Ecology Institute of the University of Potsdam. Rodents were euthanized through cervical dislocation without sedation, which is standard for small birds and rodents < 150 g body mass, in accordance with German Regulations on the protection of animals used for scientific purposes (TierSchVersV, Appendix 2). Before dissection, weights (Table A9) and morphological features of the dead rodent were recorded and a thorough collection of ectoparasites was performed for a total of five minutes using a lice comb and tweezers, conserved in 0.5 ml Eppendorf tubes filled with Ethanol 70%. Ectoparasites were assigned to three functional (and taxonomic) groups: ticks, lice, and fleas, using a stereo microscope. Rodent dissections were performed according to the protocol described in Herbreteau et al.68, for field and laboratory rodent studies. Several organs were collected and stored for PanEuropean collaborative studies on SARS-CoV-2 distribution in rodents69 and other rodent-borne diseases70.The spleen is a phagocytic filter that removes bacteria from the bloodstream and it is an antibody-producing organ71 so that recent bacterial infections can be detected. Spleen samples were kept in Ethanol 96% at 4 °C until analysis. To identify current or very recent bacterial infections, a 16 S rRNA gene amplicon sequencing was employed on rodent spleens. DNA extraction, PCR amplification, and other necessary steps were followed as detailed in Galan et al.72. In brief, each DNA extraction was analysed in two independent replicates. Three MiSeq sequencing runs were performed and the raw data are publicly available. Bacterial taxa (or parasitic taxa including Sarcocystidae) are reported as clusters, or operational taxonomic units (OTUs), are sequences that share enough similarity in the molecular level to be sorted together. OTUs were taxonomically classified using Basic Local Alignment Search Tool (BLAST) and the Silva database v138.1 to infer species or genus identity where possible.Data and statistics To investigate effects of behaviour on pathogen occurrence, we first had to identify behavioural variables that quantified consistent, inter-individual differences in behaviour7374. For this we calculated a repeatability score R for each separate variable obtained in the behavioural tests in a mixed-effect models framework75, where a value R is obtained as the proportion of the total variance accounted for by differences among individuals and the total variation, with the package rptR75. Repeatability analyses were based on 186 individuals tested in the behavioural tests, with 57 of those individuals being tested repeatedly (20 C. glareolus, 22 A.agrarius, 15 A. flavicollis). Models calculating repeatabilities always had the same structure, with the respective behavioural variable measured in the behavioural test as a response variable and only including random intercepts (individual ID) but no additional fixed factors. The appropriated error structures were calculated by considering the respective underlying distribution of each behavioural variable (Table A1). The higher the R and the more skewed away from 0 the confidence interval was, the more repeatable the particular behaviour.C. glareolus showed the highest amount of repeatable behaviour, while in both Apodemus species most behavioural variables were not repeatable (Table A1). Only the “proportion of sections” covered in the open field test was repeatable for all three species. Thus, we refrained from using combined scores as a quantitative measure of personality traits, instead we used the “proportion of sections covered” directly as an indicator of individual activity levels. Further, since the number of individuals with repeated tests was rather small, we did not calculate a personality score per animal that would account for learning or habituation in repeated testing, but we used the values obtained in the first testing round of each animal. The latency to stick the head out of the tube, “latency head” was used as a quantitative measure for boldness, even though it was repeatable only for C. glareolus in this study (Table A1), but in many earlier studies with larger sample sizes, this variable has been shown to be repeatable also for different Apodemus species197677. Due to its bimodal distribution (not sticking the head out at all, versus sticking the head out during the test) it was converted into a binary variable and called “emergence”. Binary variables showed the same repeatability patterns as the original variables (Table A1). The proportion of sections covered and emergence negatively correlated (Spearman correlation; S = 3285817, p = 0.001, Rho = −0.20; Fig. 1), but weakly enough7879, to use both variables as covariates in subsequent statistical models. Preliminary analysis (supplement tables A4 – A8) showed no difference between the two Apodemus species, neither for behavioural variables nor for the effect of behavioural variables on pathogen and ectoparasite richness, community composition and occurrence, but both mice species differed from C. glareolus. Repeatabilities of behavioural variables differed between the single Apodemus species compared to the data set of both Apodemus species combined (Table A1). These differences are most likely due to the differences in sample size and number of repeatedly tested individuals, as both crucially influence the repeatability estimations80. Based on these findings, and considering that pooling the data on mice would allow to even out the sample size differences between species, data from the two Apodemus species was combined for all subsequent analyses resulting in the presented comparisons referring to the genus, rather than the species level.To understand how sampling season, site, genus, sex and test repeat contributed to the variance observed in the chosen behavioural variables, we used them as fixed factors in generalised linear mixed models (GLMM) with either emergence or proportion of sections, as a response variable. Individual ID was always incorporated as a random factor. For both variables mixed models including all individuals tested (n = 186) were calculated considering a binomial distribution with a logit link function to model the appropriate error structure of the data. GLMMs were calculated with the package lme481.Alpha diversity of pathogens and ectoparasites within animals was quantified using richness. We calculated two different models, one only including data from pathogens found in the spleen and one considering data on pathogens and ectoparasites together. Both models were based on 93 individuals that had data for both, the behavioural test and the pathogens and ectoparasites. We chose to calculate these two separate models because the level of accuracy and repeatability of pathogen prevalence in the spleen is probably higher than of the ectoparasite counts that are extremely variable in time. We refrained from doing a separate analysis on just the ectoparasites alone, since only two taxa with sufficient data were obtained. Further, since some of the ectoparasites are transmitting the pathogens we investigated, analysing them together seemed reasonable.To analyse if richness was affected by sampling time, sampling site, genus, sex or the behavioural measures, we incorporated them as fixed factors into a GLM, with either pathogen richness, or combined pathogen and ectoparasite richness as the response variables (Table 3). Since richness is a count variable, we fitted statistical models with poisson distributions.Community composition analyses (Permutations and ordinations) required the removal of data from samples without any pathogen detection, and the removal of pathogens with less than 10% of cases infected. Thus, 71 individual samples were included to the analyses on pathogen communities (35 C. glareolus, 14 A. agrarius and 22 A. flavicollis).To quantify the variance explained by behaviour on the community compositions of pathogens, we used permutational multivariate ANOVA (adonis2 command) in the vegan package82 adding the two behavioural measures, sampling time, site, genus and sex as explanatory fixed factors. Three different permutational multivariate ANOVAs were conducted. The first one was run with data on pathogens and ectoparasites found in mice and voles, while the other two analyses were based on either data for just voles or just mice respectively. This subdivision of the data was done because mice and voles differed in the occurrence of pathogens, with some pathogens being sufficiently present (> 10% of animals infected) in only one of the genera but not the other (Tables 2 and 3). Thereby subsetting the data allowed to look at genera specific patterns of pathogen communities and how they are influenced by behavioural measures, sampling time, site and sex.To test the effects of sampling time, sampling site, genus, sex or the behavioural measures on single pathogen and ectoparasite occurrences, we used them as fixed factors in separate GLMs for each pathogen/ectoparasite type (Table 4). Occurrence (yes/no) of the respective pathogen and ectoparasite was used as binary response variable and a binomial distribution was assumed for all occurrence GLMs. In each analysis we always included the two behavioural variables as fixed factors, to test our behaviour related hypotheses. We challenged non-target covariates, i. e. genus, site, sex and season. If a covariate was improving the model fit (AIC comparison, delta AIC > 2), or if its effect on the respective response was significant, we kept it in the model, otherwise we removed it.Ethics declarationsCollection of animals was permitted by the Landesamt für Umwelt (LFU-N4-4730/11 + 10#120786/2021), capture and testing methods were permitted by the Landesamt für Arbeitsschutz, Verbraucherschutz und Gesundheit (2347-A-16-1-2020). The study complies with the applicable international, national, and/or institutional guidelines for the use of animals and with the ASAB/ABS Guidelines for the Use of Animals in Research and authors complied with the ARRIVE guidelines.

Data availability

Data from the thee MiSeq sequencing runs is available on Zenodo ([https://doi.org/10.5281/zenodo.12518285]: Run14, Run186 and Run2018).The datasets for the behavioural analyses are available from the corresponding author on reasonable request.
ReferencesJones, K. E. et al. Global trends in emerging infectious diseases. Nature 451, 990–993 (2008).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Hillegass, M. A., Waterman, J. M. & Roth, J. D. The influence of sex and sociality on parasite loads in an African ground squirrel. Behav. Ecol. 19, 1006–1011 (2008).
Google Scholar 
Webber, Q. M. R. & Willis, C. K. R. Personality affects dynamics of an experimental pathogen in little brown bats. R. Soc. Open Sci. 7, 200770 (2020).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Bohn, S. J. et al. Personality predicts ectoparasite abundance in an asocial sciurid. Ethology 123, 761–771 (2017).
Google Scholar 
Santicchia, F. et al. The price of being bold? Relationship between personality and endoparasitic infection in a tree squirrel. Mamm. Biol. 97, 1–8 (2019).
Google Scholar 
Vanden Broecke, B. et al. Relationship between population density and viral infection: A role for personality?. Ecol. Evol. 9, 10213–10224 (2019).PubMed 
PubMed Central 

Google Scholar 
Gatkowska, J., Wieczorek, M., Dziadek, B., Dzitko, K. & Dlugonska, H. Behavioral changes in mice caused by Toxoplasma gondii invasion of brain. Parasitol. Res. 111, 53–58 (2012).PubMed 
PubMed Central 

Google Scholar 
Goblirsch, M., Huang, Z. Y. & Spivak, M. Physiological and behavioral changes in honey bees (Apis mellifera) induced by Nosema ceranae infection. PLoS One 8, e58165 (2013).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
DiRienzo, N., Niemelä, P. T., Skog, A., Vainikka, A. & Kortet, R. Juvenile pathogen exposure affects the presence of personality in adult field crickets. Front. Ecol. Evol. 3, 36 (2015).
Google Scholar 
Bell, A. M. Animal personalities. Nature 447, 539–540 (2007).ADS 
CAS 
PubMed 

Google Scholar 
Biro, P. A. & Stamps, J. A. Are animal personality traits linked to life-history productivity?. Trends Ecol. Evol. 23, 361–368 (2008).PubMed 

Google Scholar 
Carter, A. J., Feeney, W. E., Marshall, H. H., Cowlishaw, G. & Heinsohn, R. Animal personality: What are behavioural ecologists measuring?. Biol. Rev. Camb. Philos. Soc. 88, 465–475 (2013).PubMed 

Google Scholar 
Réale, D., Reader, S. M., Sol, D., McDougall, P. T. & Dingemanse, N. J. Integrating animal temperament within ecology and evolution. Biol. Rev. Camb. Philos. Soc. 82, 291–318 (2007).PubMed 

Google Scholar 
Sih, A. et al. Animal personality and state–behaviour feedbacks: A review and guide for empiricists. Trends Ecol. Evol. 30, 50–60 (2015).PubMed 

Google Scholar 
VanderWaal, K. L. & Ezenwa, V. O. Heterogeneity in pathogen transmission: Mechanisms and methodology. Funct. Ecol. 30, 1606–1622 (2016).
Google Scholar 
Gao, Z. C. et al. A systematic review of the relationship between host personality traits and parasitic infection. Biology 15, 490 (2026).PubMed 
PubMed Central 

Google Scholar 
Natoli, E. et al. Bold attitude makes male urban feral domestic cats more vulnerable to Feline Immunodeficiency Virus. Neurosci. Biobehav. Rev. 29, 151–157 (2005).PubMed 

Google Scholar 
Schirmer, A., Herde, A., Eccard, J. A. & Dammhahn, M. Individuals in space: Personality-dependent space use, movement and microhabitat use facilitate individual spatial niche specialization. Oecologia 189, 647–660 (2019).ADS 
PubMed 
PubMed Central 

Google Scholar 
Schirmer, A., Hoffmann, J., Eccard, J. A. & Dammhahn, M. My niche: individual spatial niche specialization affects within- and between-species interactions. Proc. Royal Soc. B: Biol. Sci. 287, 20192211 (2020).
Google Scholar 
Wanelik, K. M. et al. Superspreaders have lower gut microbial alpha-diversity and distinct gut microbial composition in a natural rodent population. Anim. Microbiome. 7, 42 (2025).CAS 
PubMed 
PubMed Central 

Google Scholar 
Petkova, I., Abbey-Lee, R. N. & Løvlie, H. Parasite infection and host personality: Glugea-infected three-spined sticklebacks are more social. Behav. Ecol. Sociobiol. 72, 173 (2018).PubMed 
PubMed Central 

Google Scholar 
Dizney, L. & Dearing, M. D. The role of behavioural heterogeneity on infection patterns: Implications for pathogen transmission. Anim. Behav. 86, 911–916 (2013).
Google Scholar 
Bitam, I., Dittmar, K., Parola, P., Whiting, M. F. & Raoult, D. Fleas and flea-borne diseases. Int. J. Infect. Dis. 14, e667–e676 (2010).PubMed 

Google Scholar 
Madison-Antenucci, S., Kramer, L. D., Gebhardt, L. L. & Kauffman, E. Emerging tick-borne diseases. Clin. Microbiol. Rev. 33, 10.1128/cmr.00083 − 18 (2020).
Google Scholar 
Erixon, F., Eccard, J. A., Huneke, R. & Dammhahn, M. A behavioral syndrome of competitiveness in a non-social rodent. Behav. Ecol. Sociobiol. 78, 98 (2024).
Google Scholar 
Gyuris, E., Hankó, J. F., Feró, O. & Barta, Z. Personality and ectoparasitic mites (Hemipteroseius adleri) in firebugs (Pyrrhocoris apterus). Behav. Process. 122, 67–74 (2016).
Google Scholar 
Arnold, W. & Anja, V. L. Ectoparasite loads decrease the fitness of alpine marmots (Marmota marmota) but are not a cost of sociality. Behav. Ecol. 4, 36–39 (1993).
Google Scholar 
Fitze, P. S., Tschirren, B. & Richner, H. Life history and fitness consequences of ectoparasites. J. Anim. Ecol. 73, 216–226 (2004).
Google Scholar 
Maaz, D. et al. Factors associated with diversity, quantity and zoonotic potential of ectoparasites on urban mice and voles. PLOS ONE. 13, e0199385 (2018).PubMed 
PubMed Central 

Google Scholar 
Wat, K. K. Y., Herath, A. P. H. M., Rus, A. I., Banks, P. B. & Mcarthur, C. Space use by animals on the urban fringe: Interactive effects of sex and personality. Behav. Ecol. 31, 330–339 (2020).
Google Scholar 
Wauters, L. A. et al. Personality traits, sex and food abundance shape space use in an arboreal mammal. Oecologia 196, 65–76 (2021).ADS 
PubMed 
PubMed Central 

Google Scholar 
Balashov, Y. S. Types of parasitism of acarines and insects on terrestrial vertebrates. Entmol. Rev. 86, 957–971 (2006).
Google Scholar 
Han, B. A., Schmidt, J. P., Bowden, S. E. & Drake, J. M. Rodent reservoirs of future zoonotic diseases. Proc. Natl. Acad. Sci. U. S. A. 112, 7039–7044 (2015).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Gutiérrez, R. et al. Bartonella infection in rodents and their flea ectoparasites: An overview. Vector Borne Zoonotic Dis. 15, 27–39 (2015).PubMed 
PubMed Central 

Google Scholar 
Silaghi, C., Pfeffer, M., Kiefer, D., Kiefer, M. & Obiegala, A. Bartonella, rodents, fleas and ticks: A molecular field study on host-vector-pathogen associations in Saxony, Eastern Germany. Microb. Ecol. 72, 965–974 (2016).ADS 
PubMed 

Google Scholar 
Spitalská, E., Boldis, V., Kostanová, Z., Kocianová, E. & Stefanidesová, K. Incidence of various tick-borne microorganisms in rodents and ticks of central Slovakia. Acta Virol. 52, 175–179 (2008).PubMed 

Google Scholar 
Gautier-Bouchardon, A. V. Antimicrobial resistance in Mycoplasma spp. Microbiol. Spectr. 6, 10.1128/microbiolspec.arba-0030–2018 (2018).PubMed 
PubMed Central 

Google Scholar 
Abbate, J. L. et al. Pathogen community composition and co-infection patterns in a wild community of rodents. Peer Community J. https://doi.org/10.24072/pcjournal.370 (2024).Article 

Google Scholar 
Razzauti, M. et al. A comparison between transcriptome sequencing and 16S metagenomics for detection of bacterial pathogens in wildlife. PLoS Negl. Trop. Dis. 9, e0003929 (2015).PubMed 
PubMed Central 

Google Scholar 
Harasawa, R., Fujita, H., Kadosaka, T., Ando, S. & Rikihisa, Y. Proposal for ‘CandidatusMycoplasma haemomuris subsp. musculi’ in mice, and ‘CandidatusMycoplasma haemomuris subsp. ratti’ in rats. Int. J. Syst. Evol. Microbiol. 65, 734–737 (2015).CAS 
PubMed 

Google Scholar 
Otake, S., Dee, S., Corzo, C., Oliveira, S. & Deen, J. Long-distance airborne transport of infectious PRRSV and Mycoplasma hyopneumoniae from a swine population infected with multiple viral variants. Vet. Microbiol. 145, 198–208 (2010).PubMed 

Google Scholar 
Obiegala, A. et al. CandidatusNeoehrlichia mikurensis and Anaplasma phagocytophilum: Prevalences and investigations on a new transmission path in small mammals and ixodid ticks. Parasit. Vectors 7, 563 (2014).PubMed 
PubMed Central 

Google Scholar 
Portillo, A., Santibáñez, P., Palomar, A. M., Santibáñez, S. & Oteo, J. A. ‘CandidatusNeoehrlichia mikurensis’ in Europe. New Microbes New Infect. 22, 30–36 (2018).CAS 
PubMed 
PubMed Central 

Google Scholar 
Silaghi, C., Beck, R., Oteo, J. A., Pfeffer, M. & Sprong, H. Neoehrlichiosis: An emerging tick-borne zoonosis caused by CandidatusNeoehrlichia mikurensis. Exp. Appl. Acarol. 68, 279–297 (2016).PubMed 

Google Scholar 
Wennerås, C. Infections with the tick-borne bacterium CandidatusNeoehrlichia mikurensis. Clin. Microbiol. Infect. 21, 621–630 (2015).PubMed 

Google Scholar 
Lipatova, I. et al. Bartonella infection in small mammals and their ectoparasites in Lithuania. Microbes Infect. 17, 884–888 (2015).CAS 
PubMed 

Google Scholar 
Goldberg, A. R., Conway, C. J. & Biggins, D. E. Flea sharing among sympatric rodent hosts: Implications for potential plague effects on a threatened sciurid. Ecosphere 11, e03033 (2020).
Google Scholar 
Koprivnikar, J., Gibson, C. H. & Redfern, J. C. Infectious personalities: Behavioural syndromes and disease risk in larval amphibians. Proc. Biol. Sci. 279, 1544–1550 (2011).PubMed 
PubMed Central 

Google Scholar 
Rollins, R. E. et al. Repeatable differences in exploratory behaviour predict tick infestation probability in wild great tits. 03.06.978973 Preprint at (2020). https://doi.org/10.1101/2020.03.06.978973 (2021).Mejlon, H. A. & Jaenson, T. G. T. Questing behaviour of Ixodes ricinus ticks (Acari: Ixodidae). Exp. Appl. Acarol. 21, 747–754 (1997).
Google Scholar 
Aplin, L. M. et al. Individual personalities predict social behaviour in wild networks of great tits (Parus major). Ecol. Lett. 16, 1365–1372 (2013).CAS 
PubMed 

Google Scholar 
Payne, E. et al. Personality, space use, and networks directly and indirectly explain tick infestation in a wild population of lizards. Ecol. Monogr. 95, e70000 (2025).
Google Scholar 
Tack, W., Madder, M., Baeten, L., Frenne, P. D. & Verheyen, K. The abundance of Ixodes ricinus ticks depends on tree species composition and shrub cover. Parasitology 139, 1273–1281 (2012).ADS 
CAS 
PubMed 

Google Scholar 
Eccard, J. A. et al. Fitness, risk taking, and spatial behavior covary with boldness in experimental vole populations. Ecol. Evol. 12, e8521 (2022).PubMed 
PubMed Central 

Google Scholar 
Zuk, M. & McKean, K. A. Sex differences in parasite infections: Patterns and processes. Int. J. Parasitol. 26, 1009–1024 (1996).CAS 
PubMed 

Google Scholar 
Webber, Q. M. R., McGuire, L. P., Smith, S. B. & Willis, C. K. R. Host behaviour, age and sex correlate with ectoparasite prevalence and intensity in a colonial mammal, the little brown bat. Behaviour 152, 83–105 (2015).
Google Scholar 
Giles, N. Behavioural effects of the parasite Schistocephalus solidus (Cestoda) on an intermediate host, the three-spined stickleback, Gasterosteus aculeatus L. Anim. Behav. 31, 1192–1194 (1983).
Google Scholar 
Hechtel, L. J., Johnson, C. L. & Juliano, S. A. Modification of Antipredator Behavior of Caecidotea Intermedius by Its Parasite Acanthocephalus Dirus. Ecology 74, 710–713 (1993).
Google Scholar 
Baldauf, S. A. et al. Infection with an acanthocephalan manipulates an amphipod’s reaction to a fish predator’s odours. Int. J. Parasitol. 37, 61–65 (2007).PubMed 

Google Scholar 
Lafferty, K. D. & Morris, A. K. Altered Behavior of Parasitized Killifish Increases Susceptibility to Predation by Bird Final Hosts. Ecology 77, 1390–1397 (1996).
Google Scholar 
Hay, J., Hutchison, W. M., Aitken, P. P. & Graham, D. I. The effect of congenital and adult-acquired Toxoplasma infections on activity and responsiveness to novel stimulation in mice. Ann. Trop. Med. Parasitol. 77, 483–495 (1983).CAS 
PubMed 

Google Scholar 
Webster, J. P. The effect of Toxoplasma gondii and other parasites on activity levels in wild and hybrid Rattus norvegicus. Parasitology 109, 583–589 (1994).ADS 
PubMed 

Google Scholar 
Firozpoor, J., Gardini, R., Huezo, M. E. E. & Eccard, J. A. Rodent species composition in urban and forested areas in eastern Germany. Biodiversity Data Journal 13, e143224 (2025).PubMed 
PubMed Central 

Google Scholar 
Eccard, J. A. & Klemme, I. Reducing mortality of shrews in rodent live trapping—A method increasing live-trap selectivity with shrew exits. Ann. Zool. Fenn. 50, 371–376 (2013).
Google Scholar 
Herde, A. & Eccard, J. A. Consistency in boldness, activity and exploration at different stages of life. BMC Ecology 13, 49 (2013).PubMed 
PubMed Central 

Google Scholar 
Archer, J. Tests for emotionality in rats and mice: A review. Anim. Behav. 21, 205–235 (1973).CAS 
PubMed 

Google Scholar 
Gould, T. D., Dao, D. T. & Kovacsics, C. E. The Open Field Test In (ed. Gould, T. D.) (2009).Herbreteau, V. et al. Protocols for Field and Laboratory Rodent Studies Vol. 5, p. 46 (p. (Kasetsart University, 2011).Bourret, V. et al. Serologic Surveillance for SARS-CoV-2 Infection among Wild Rodents, Europe – 28, Number 12—December 2022 – Emerging Infectious Diseases journal – CDC. (2022). https://doi.org/10.3201/eid2812.221235 doi:10.3201/eid2812.221235.Sluydts, V. et al. Drivers of host-pathogen community assemblies in European forests and urban green spaces. 09.19.613856 Preprint at (2024). https://doi.org/10.1101/2024.09.19.613856 (2024).Bohnsack, J. F. & Brown, E. J. The role of the spleen in resistance to infection. Annu Rev Med 37, 49–59 (1986).CAS 
PubMed 

Google Scholar 
Galan, M. et al. 16S rRNA amplicon sequencing for epidemiological surveys of bacteria in wildlife. mSystems 1, 101128msystems00032–101128msystems00016 (2016).
Google Scholar 
Réale, D. et al. Personality and the emergence of the pace-of-life syndrome concept at the population level. Philosophical Transactions of the Royal Society B: Biological Sciences 365, 4051–4063 (2010).
Google Scholar 
Bell, A. M., Hankison, S. J. & Laskowski, K. L. The repeatability of behaviour: A meta-analysis. Anim Behav 77, 771–783 (2009).PubMed 
PubMed Central 

Google Scholar 
Stoffel, M. A., Nakagawa, S. & Schielzeth, H. rptR: Repeatability estimation and variance decomposition by generalized linear mixed-effects models. Methods Ecol Evol 8, 1639–1644 (2017).
Google Scholar 
Dammhahn, M., Mazza, V., Schirmer, A., Göttsche, C. & Eccard, J. A. Of city and village mice: Behavioural adjustments of striped field mice to urban environments. Scientific Reports 10, 13056 (2020).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Eccard, J. A., Mazza, V., Holland, C. & Stuart, P. The timid invasion: Behavioural adjustments and range expansion in a non-native rodent. Proceedings of the Royal Society B: Biological Sciences 290, 20230823 (2023).
Google Scholar 
Dormann, C. F. et al. Collinearity: A review of methods to deal with it and a simulation study evaluating their performance. Ecography 36, 27–46 (2013).ADS 

Google Scholar 
Harrison, X. A. et al. A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ 6, e4794 (2018).PubMed 
PubMed Central 

Google Scholar 
Wolak, M. E., Fairbairn, D. J. & Paulsen, Y. R. Guidelines for estimating repeatability. Methods in Ecology and Evolution 3, 129–137 (2012).
Google Scholar 
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67, 1–48 (2015).
Google Scholar 
Oksanen, J. et al. Package’vegan’. The comprehensive R archive network. (2020).Download referencesAcknowledgementsWe are grateful to Riccardo Gardini (University of Modena and Reggio Emilia), who joined our team as a research student and Dr. Annika Schirmer, (University of Potsdam), for her support and assistance. We also would like to thank the GenSeq technical facilities of MEEB (CNRS and University of Montpellier) hosted by ISEM (CNRS, University of Montpellier and IRD) and Audrey Weber (INRAE-AGAP) for the MiSeq sequencing runs, and the Genotoul bioinformatics platform Toulouse Midi-Pyrénées (Bioinfo Genotoul).FundingOpen Access funding enabled and organized by Projekt DEAL. Our research was funded through the European H2020 (WP 2018–2020) call and the 2018–2019 BiodivERsA joint call for research proposals, under the BiodivErsA3 ERA-Net COFUND program and cofunded by Agence Nationale de la Recherche and the German Science foundation (DFG, Projektnummer 428675001).Author informationAuthors and AffiliationsAnimal Ecology, Institute for Biochemistry and Biology, University of Potsdam, Potsdam, GermanyJana A. Eccard, Jasmin Firozpoor & Mario EscobarCBGP, INRAE, CIRAD, IRD, Institut Agro, Univ Montpellier, Montpellier, FranceMaxime Galan & Nathalie CharbonnelAuthorsJana A. EccardView author publicationsSearch author on:PubMed Google ScholarJasmin FirozpoorView author publicationsSearch author on:PubMed Google ScholarMario EscobarView author publicationsSearch author on:PubMed Google ScholarMaxime GalanView author publicationsSearch author on:PubMed Google ScholarNathalie CharbonnelView author publicationsSearch author on:PubMed Google ScholarContributionsJAE: Conceptualization, Methodology, Validation, Formal Analysis, Funding Acquisition, Investigation, Resources, Data Curation, Project administration, Supervision, Writing – Original Draft, Writing – Review & Editing, Visualization. JF: Conceptualization, Methodology, Investigation, Data Curation, Project administration, Writing – Review & Editing. ME: Investigation, Methodology, Writing – Review & Editing. MG: Investigation, Methodology, Writing – Review & Editing. NC: Funding Acquisition, Investigation, Methodology, Project administration, Writing – Review & Editing.Corresponding authorCorrespondence to
Jana A. Eccard.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary InformationBelow is the link to the electronic supplementary material.Supplementary Material 1 (download PDF )Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and permissionsAbout this articleCite this articleA. Eccard, J., Firozpoor, J., Escobar, M. et al. Individual activity of forest rodents correlates to pathogen communities.
Sci Rep 16, 14684 (2026). https://doi.org/10.1038/s41598-026-51276-6Download citationReceived: 28 November 2025Accepted: 27 April 2026Published: 09 May 2026Version of record: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51276-6Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
Keywords
Bartonella

Borrelia

Mycoplasma

Sarcocystidae

Apodemus

Clethrionomys

Myodes
BehaviourAnimal personalityActivity More

AbstractLivestock systems contribute substantially to Greenhouse Gas (GHG) emissions, highlighting the need for interventions that enhance productivity while lowering environmental impact. This study assessed GHG emissions and emission intensity (EI) in Ethiopian Menz and Bonga sheep managed under Community-Based Breeding Programs (CBBP) compared with non-CBBP flocks. Using FAO’s GLEAM-i model with long-term performance data (2009–2022) and surveys of 321 households, we found that CBBPs reduced total emissions by 14.56% in Menz and 7.04% in Bonga sheep, while protein output rose by 42% and 2%, respectively. EI declined by 21.49% in Menz and 6.29% in Bonga. At the household level, CBBP flocks achieved markedly lower EI, 39.39% in Menz and 30.68% in Bonga, despite higher absolute emissions from larger flocks. Methane, mainly from enteric fermentation, accounted for a higher proportion compared to other GHG gases. EI was positively associated with reproductive inefficiencies while negatively associated with growth traits, underscoring key levers for mitigation. These results demonstrate that CBBPs deliver both productivity gains and environmental co-benefits, offering a scalable model for climate-smart small ruminant development with relevance beyond Ethiopia.

AcknowledgementsWe sincerely acknowledge DBARC and BARC for providing primary data from CBBP villages. We extend our gratitude to the International Center for Agricultural Research in the Dry Areas (ICARDA) and AbacusBio for the development of data collection tools, Digital Tool for Recording and Evaluation of Animals (DTREO), which improved data quality and accessibility. The Australian Center for International Agricultural Research (ACIAR) and the Unique Land Use project are thanked for their funding support of the first author as part of his PhD research. We appreciate the efforts of researchers and enumerators for their contributions to data collection, expert elicitation, and the provision of valuable insights into the two sheep CBBPs. Special thanks to Mr. Asfaw Bisrat and Metsafe Mamiru for coordinating fieldwork and managing households for primary data collections. We also acknowledge the CBBP enumerators and cooperative leaders for their role in ensuring high-quality data collection and sustaining village operations.FundingThis work did not receive any specific grant from funding agencies.Author informationAuthors and AffiliationsCollege of Agriculture and Environmental Sciences, Bahir Dar University, Bahir Dar, EthiopiaAssemu Tesfa, Mengistie Taye, Aynalem Haile & Zerihun NigussieAndassa Livestock Research Center, Amhara Agricultural Research Institute, Bahir Dar, EthiopiaAssemu TesfaInstitute of Biotechnology Research, Bahir Dar University, Bahir Dar, EthiopiaMengistie TayeInternational Centre for Agricultural Research in the Dry Areas, Addis Ababa, EthiopiaAynalem Haile & Tesfaye GetachewInternational Centre for Agricultural Research in the Dry Areas, Rabat, MoroccoDina NajjarUnique Land Use GmbH, Addis Ababa, EthiopiaShigdaf MekuriawGlobal Engagement Office, College of Agricultural and Environmental Sciences, University of California, Davis, 95616, USAShimels E. WassieDebre Birhan Agricultural Research Center, Amhara Agricultural Research Institute, Debre Birhan, EthiopiaShanbel Besufkad & Checkol DemisSowthwest Ethiopia Agricultural Research Institute, Bonga, EthiopiaZelalem AbateUnique Land Use GmbH, Freiburg, GermanySuzanne van Dijk & Andreas WilkesInternational Livestock Research Institute, Addis Ababa, EthiopiaDawit SolomonAuthorsAssemu TesfaView author publicationsSearch author on:PubMed Google ScholarMengistie TayeView author publicationsSearch author on:PubMed Google ScholarAynalem HaileView author publicationsSearch author on:PubMed Google ScholarZerihun NigussieView author publicationsSearch author on:PubMed Google ScholarDina NajjarView author publicationsSearch author on:PubMed Google ScholarShigdaf MekuriawView author publicationsSearch author on:PubMed Google ScholarShimels E. WassieView author publicationsSearch author on:PubMed Google ScholarShanbel BesufkadView author publicationsSearch author on:PubMed Google ScholarZelalem AbateView author publicationsSearch author on:PubMed Google ScholarCheckol DemisView author publicationsSearch author on:PubMed Google ScholarTesfaye GetachewView author publicationsSearch author on:PubMed Google ScholarSuzanne van DijkView author publicationsSearch author on:PubMed Google ScholarAndreas WilkesView author publicationsSearch author on:PubMed Google ScholarDawit SolomonView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
Assemu Tesfa.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleTesfa, A., Taye, M., Haile, A. et al. Evaluation of greenhouse gas emissions from Bonga and Menz sheep breeds managed under community-based breeding program.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-39912-7Download citationReceived: 29 September 2025Accepted: 09 February 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-39912-7Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsCommunity-based breedingEmission intensityEnvironmental footprintGreenhouse gasGLEAM-iSelection efficiency More

AbstractThe European adder (Vipera berus), classified as critically endangered in Germany, faces severe threats from habitat loss, particularly the decline of primary habitats such as peatlands. Peatland restoration is essential but may pose risks when hibernacula are not considered during rewetting, potentially leading to flooding, site loss, and population declines. Similar risks arise from construction activities in secondary habitats. Protecting these key sites is therefore vital for species conservation but requires reliable methods to locate them. Traditional monitoring approaches, such as visual surveys and telemetry, are often invasive, imprecise, or resource-intensive. This study tested whether a trained adder detection dog could identify adder scent—using sheds and fecal swabs—beneath substrate in controlled line-up experiments and under field conditions. In the line-up tests, we assessed how sample type, burial depth, and substrate influenced detection performance, whereas field searches evaluated detection success under more realistic conditions. In a pilot study, we further examined whether the dog could generalize from collected odor samples to live adders by testing its ability to locate the hibernacula of two previously tagged individuals. In the line-up experiments, the dog identified both sheds and fecal swabs with high sensitivity, performing slightly better with sheds. Detection remained effective at depths of up to 1.2 m, although detection success declined with increasing depth. Under field conditions, the dog reliably detected samples buried at a depth of 50 cm without any false indications. In the pilot study, the dog successfully located hibernating adders, demonstrating that training non-invasively with collected odor samples rather than live animals can be sufficient for this purpose. These findings suggest that detection dogs represent a promising, non-invasive, and efficient tool for locating adder hibernacula. Future studies should include multiple dogs, larger sample sizes, and additional tagged individuals to further validate detection performance under field conditions.

IntroductionGlobal biodiversity is declining rapidly, with 19.4% of species in Europe at risk of extinction, including 18% of vertebrates1. This loss is primarily driven by land-use change and climate change1. Reptiles are particularly affected, with 21.1% (1,829 species) of the nearly 10,000 known species currently considered threatened worldwide2. This development even affects reptile species with formerly wide distributions, such as the European adder (Vipera berus)3. Across large parts of Europe, particularly in Western Europe, the species has experienced pronounced population declines4,5,6. In Germany, adder populations have decreased by an estimated 50–70% over the past 100 years; the species is therefore classified as highly endangered nationwide and, in some federal states including North Rhine-Westphalia, as threatened with extinction, with continuing negative population trends4,7. As a predator, it also fulfills a functional role within ecosystems, meaning that population declines may have ecological consequences beyond the species itself3. The main drivers of this decline are habitat loss and increasing habitat fragmentation7,8. Due to its limited dispersal ability and high site fidelity, the adder is particularly sensitive to habitat alteration, as habitat fragmentation rapidly leads to small, isolated populations with an increased risk of inbreeding and local extinctions3,4. This highlights the importance of conserving existing habitats and structures, especially peatlands, which constitute key habitats for adders but have been severely reduced and degraded over recent decades7,9,10. While peatland restoration is an essential conservation measure, it also entails risks7,11. Inappropriate renaturation practices, particularly uncontrolled increases in water levels, can destroy critical hibernacula, which are often used repeatedly by adders, leading to high mortality and local population collapses7,11,12. Similar risks also arise in secondary habitats, for example during construction or habitat management activities7. Adders hibernate in protected sites such as root stumps, embankments, and burrows, with depths ranging from 5 cm to 1.25 m7,10. They are very reluctant to use newly created hibernacula, making preservation of existing ones essential11. Traditional methods for locating hibernacula are often inefficient and time-consuming, as they mainly rely on rough estimates from basking site observations and are therefore imprecise7,10,12. Telemetric surveys allow accurate localization but involve relatively high costs, cause significant disturbance to the animals, and risk missing untagged individuals using different hibernacula. Non-invasive, cost-effective methods are needed for more efficient hibernacula detection. One solution could be the use of wildlife detection dogs. Studies show that dogs outperform humans in 88.7% of cases, detecting targets faster and more accurately, even in dense vegetation13,14,15. While mammals and birds are up to now relatively well studied using detection dogs, reptiles remain underrepresented14. Even though some research has demonstrated their ability to detect snakes, lizards, or tortoises15,16,17, studies focusing on reptiles occupying underground refuges are still scarce. Existing work on tortoises and Indigosnakes in burrows has shown promising results15,18, yet it remains largely unclear how detectability is affected by burial depth and substrate type. Environmental factors such as wind, temperature, and humidity may further influence detection success19,20. Although dogs have been shown to detect amphibians beneath a 20 cm soil layer21, the applicability of this approach to deeper underground reptile refuges has not been evaluated. As a result, it remains unclear whether this approach also works for reptiles that are in deeper hibernation, such as the European adder.This study evaluates the feasibility of using detection dogs to locate adder hibernacula by assessing odor detectability under controlled conditions and detection performance in the field. Specifically, it tests whether a trained adder detection dog can reliably detect adder odor in standardized line-up tests across varying depths and substrates, and examines how sample type, substrate type, and depth affect detection success. The study further evaluates odor detection beneath substrate under different field conditions and assesses whether the dog can generalize from training samples to live adders and accurately locate them in the field.Materials and methodsThe study was conducted in three experimental steps. These comprised controlled depth-variable line-up trials, a field experiment with double-blind searches with buried odor samples at two sites, and a pilot field experiment conducted in areas with known adder hibernation. Adder sheds and fecal swabs served as odor samples in the line-up and field experiments.Samples and storageAdder sheds were collected from both wild and captive individuals using disposable gloves to avoid contamination. All sheds were stored individually in sterile, tightly sealed glass jars. To prevent mould growth, sheds were allowed to dry completely prior to storage. Fecal samples were collected using sterile swabs during routine monitoring at the Dortmund–Ems Canal, when snakes produced fresh scat during handling. Swab samples were transported in cooled containers and stored at − 18 °C. Prior to each trial, swabs were defrosted for approximately 30 min. In total, 21 sheds and 45 fecal swabs were used across training and experimental trials. Fecal swabs were obtained from approximately seven different individual adders, resulting in a total of 45 swab samples. Sheds were collected from 21 individuals, including nine samples from three wild adder populations in Germany and twelve from individuals housed in outdoor enclosures at various zoological institutions in Germany. Before use, all sheds were visually examined and confirmed as Vipera berus based on characteristic morphological features visible in the shed skin. The individuals providing fecal swabs and sheds did not overlap. Samples used during training were not reused in subsequent experimental trials.Detection dog selection and trainingA 23-month-old male English Springer Spaniel was selected for its proven scent detection abilities. Training began at 16 weeks of age and lasted for approximately 1.5 years prior to the start of the experiments. Initial training used chamomile tea as a substitute odor before transitioning to adder scents. The dog was trained using classical and operant conditioning to indicate a target odor by lying down. Correct indications were marked with a clicker and immediately followed by delivery of a primary reward (food or play). A clicker functions as a conditioned (secondary) reinforcer that marks the precise moment of the correct response and bridges the temporal delay to reinforcement22. Training comprised searches in a standard line-up of up to ten clay pots, as well as multiple searches using a depth-adjustable line-up that was also employed in the experiments of this study (Supplementary Fig. 1). Both training approaches were conducted using the substitute odor and the target odors (adder sheds and fecal swabs). Field search training was carried out across multiple areas differing in habitat type and structural complexity, including grasslands and pastures, forested sites, and rock piles. Training scenarios included variation in target placement, pooled and blank searches and included single-blind as well as double-blind search designs. Searches at varying depths were conducted with both the substitute odor and the target odors. Both sample types were used equally throughout training to prevent bias toward a specific odor source.General experimental proceduresLine-up and field experiments were conducted using a double-blind design. In double-blind searches, the person who placed the samples or who knew the position of the search object was not present during the search to prevent potential handler bias and unintentional cueing23,24. For all experiments, the dog was handled on a 5 m leash. Latex-free, powder-free disposable gloves from different manufacturers were used for sample handling to minimize contamination. Odor samples were placed inside plastic tubes (fecal swabs: 1 cm × 3 cm; sheds: 7 cm × 2.5 cm), which served as carrier material to allow sample reuse while reducing odor contamination of the experimental setup. To enable odor release, tube openings were covered with fiberglass mesh rather than lids. Unused sterile swabs placed in identically prepared tubes were used as blanks in swab trials. In experiments using sheds, identically prepared tubes without odor material served as blanks. After initial use, swab samples were stored at 5 °C until reuse. For the line-up and field experiments, separate sample sets were used. One batch of swabs and sheds was assigned exclusively to the line-up experiments, and another batch exclusively to the field experiments. Within each batch, samples were randomly selected for individual trials. In field searches, separate sample sets were used for each site to ensure that swab samples from the embankment and the pasture were not mixed. This approach ensured consistent sample quality and prevented potential degradation caused by prior burial or prolonged exposure to ambient temperatures. If a sample used in the line-up experiments became heavily soiled and was considered unsuitable for further use (e.g., due to visible contamination with peat), it was removed from the batch and replaced to maintain a constant number of samples within the experimental set.Depth-variable line-upTo assess the effects of sample type, depth, and substrate, a depth-adjustable line-up was used (Supplementary Fig. 1). The setup included five PVC pipes in a box (120 cm high × 40 cm wide × 260 cm long), each run filled with either stones (3–10 cm) or white peat—substrates typical of adder hibernacula. Each test included one positive and four negative targets with its position chosen randomly, with 35 runs per sample/substrate type (140 runs total). Hiding depth increased by 20 cm every five runs, from surface level to 120 cm. The sample tubes were positioned such that the fiberglass side faced upward or at least diagonally upward, and never downward, to ensure that upward odor dispersal was not artificially constrained. Used substrate was replaced after each run to avoid scent contamination. Stones that had already been used once were placed in water overnight, thoroughly rinsed and air-dried afterwards. This allowed the stones to be reused the next day. Samples were exposed for at least 30 min per 20 cm of depth (up to 180 min), following Glover et al.21. The dog was given up to two opportunities to search all tubes. If an alert occurred during the first search round, the trial was terminated, and no second round was conducted. Misses and false alerts were recorded if no correct indication was made by communicating with the person that placed the samples by radio after the dog showed the indication behavior. Correct indications scored as 1; all others as 0.Field experimentTo evaluate detection success under realistic field conditions, transect searches were conducted. Searches took place on a horse pasture in Soest (North Rhine-Westphalia, Germany) and on a north-facing embankment at the Dortmund–Ems Canal near Senden (North Rhine-Westphalia, Germany); both sites were known to be free of adder presence. At each site, twelve unique transects of approximately 50 m² were established to prevent interference from relict odors. Square transects were used on the pasture, whereas elongated transects were used on the embankment to account for the linear structure and variable width of the site (mean dimensions approximately 3.5 × 15 m). All transects were searched once, with an equal number of transects per sample type. Searches were conducted between January 27 and February 10, 2024, across five days; on three days, searches were split into morning and afternoon blocks, whereas the remaining two days consisted of a single block each. Per transect, 2–4 positive samples and four blank samples were buried at approximately 50 cm depth, a depth representative of typical adder hibernacula in western Europe10, resulting in a total of 70 positive samples (35 per site). The placement of positive and blank samples followed no predefined pattern. After burial, the soil was compacted to avoid visual cues indicating sample locations. To prevent the dog from tracking the assistant’s scent, the assistant moved through each transect in a criss-cross pattern and deliberately touched different areas throughout the transect. In some locations, the soil was disturbed without placing a sample. Samples were buried approximately 75 min prior to the start of each search, following Glover et al.21. Searches were initiated by a verbal command from the handler, who was free to choose the starting position. For each alert, the time required until the indication was recorded; following documentation of the alert, the search was resumed, and time measurement was restarted. Wind speed and direction, temperature, and humidity were recorded. Upon an alert by the dog (e.g. lying down), time was stopped and GPS coordinates and photographic documentation were collected, after which the dog resumed searching. Once the handler considered the transect fully searched, the search was terminated and the next transect was visited. Detection points were subsequently compared with the actual sample locations, and correct, false, and missed indications were documented after completion of the search. All samples were removed after each transect search. To prevent frustration, the dog was rewarded after each alert.Feasibility pilot field test with live addersThe surveys were conducted in the Venner Moor near Senden (Coesfeld district, North Rhine-Westphalia, Germany) and in a second study area located along the nearby south-facing side of the Dortmund–Ems Canal; both areas are known to support relatively large adder populations. The Dortmund–Ems Canal study area comprised a south-facing stone embankment of approximately 2 km in length, adjacent wooded areas, and three substitute habitats consisting of stone bars and brushwood piles. The tests took place on January 10, 2024, under dry weather conditions, at a temperature of − 4 °C and with wind speeds of up to 6.2 m/s. In total, four transect searches were conducted. At each study area, one search was carried out in a transect with a telemetrically confirmed hibernaculum (based on telemetric monitoring data from November 2023) and one search was conducted in a transect with a high probability of containing a hibernaculum, as assessed by expert judgement. In areas with telemetrically confirmed hibernacula, the handler was assigned transects of approximately 375 m² (15 × 25 m) at the Dortmund–Ems Canal and approximately 320 m² (8 × 40 m) in the Venner Moor. The two potential hibernacula areas comprised transects of approximately 70 m² each. The confirmed transects were located in the eastern replacement habitat near the Dortmund–Ems Canal and on a peat dam in the Venner Moor. To ensure a double-blind search, the person who was aware of the telemetrically determined adder locations remained out of sight during the searches. Upon an alert, the dog was initially rewarded with a click and a ball, after which the position was checked. All alerts, including those without immediate verification, were rewarded, as the presence of additional, non-tracked individuals could not be excluded. Search data, including position, weather parameters (temperature, wind speed, wind direction, relative humidity, precipitation) and time, were recorded.Statistical analysisWe calculated sensitivity as a classical signal detection theory parameter using the formula (correct alerts / positive targets) × 100. In addition to sensitivity, effectiveness was calculated as (correct alerts / [positive targets + false alerts]) × 100 as a complementary measure of detection performance that accounts for both missed targets and false alerts following Matthew et al.25. This prevents an overestimation of performance, for example in dogs that indicate frequently but non-specifically. Both metrics were calculated for overall performance as well as separately for the different combinations of sample type and substrate.Analyses were conducted in RStudio v2023.12.026 using R statistical software v4.3.227. The effects of sample type, depth, and substrate on search success (correct = 1, missed/false = 0) in the depth line-up trials were analyzed using a binomial generalized linear model (GLM). Model selection was based on the lowest AIC, and overall model significance was assessed using likelihood-ratio (omnibus) chi-square tests comparing the full model to the null model. Effect sizes and model fit were evaluated using odds ratios and Nagelkerke’s pseudo-R², while model assumptions were examined based on deviance residuals and simulation-based residual diagnostics implemented in DHARMa28. Models were tested for overdispersion, multicollinearity, and residual autocorrelation. Multicollinearity was evaluated using variance inflation factors (VIFs), and residual autocorrelation was assessed using model-based diagnostics implemented in the R package performance29. To examine whether the effect of depth depends on substrate type and sample type, we considered models including interaction terms. However, estimating such interactions requires reliable model fitting and sufficient variation within each substrate–sample type combination to estimate separate depth effects. In our data, correct and incorrect detections were highly unbalanced within most combinations, such that interaction models could not be fitted reliably and would not yield trustworthy results.Due to quasi-complete separation of the data (69 correct detections out of 70), inferential statistical analyses of detection success were not conducted. Under such conditions, parameter estimates from generalized linear or mixed-effects models are unstable and not meaningful, regardless of model structure. Detection performance is therefore reported descriptively using sensitivity and effectiveness metrics. Inferential analyses of search time were likewise not performed, as environmental conditions were strongly structured by site and time block and did not show sufficient variation to allow reliable estimation of individual effects. Search time was therefore reported descriptively to provide an overview of the dog’s search speed at both study sites. No statistical analysis was performed for the pilot study due to limited adder sightings. All plots were generated using ggplot2 and ggpubr. Analyses were conducted using the R packages tidyverse30, car and psych31.Ethics declarationThis study was conducted in full compliance with ethical standards for wildlife research and in accordance with German and European Union regulations concerning the protection of native species. All research activities were approved by the appropriate regulatory authorities, including the Lower Nature Conservation Authority, Coesfeld District and corresponding landowners. The European adder is a protected species in Germany; therefore, all work involving capture, handling, or monitoring was carried out with minimal disturbance to the animals and their natural habitat. Trained personnel handled adders using safe and humane techniques to reduce stress and prevent injury to both animals and researchers. No animals were harmed, killed, or removed from their habitat unless specifically authorized under the issued permit. After data collection, all individuals were released at the site of capture as promptly as possible. The study design followed the ethical guidelines of the German Animal Welfare Act (Tierschutzgesetz) and the European Directive 2010/63/EU on the protection of animals used for scientific purposes, as applicable to non-invasive field studies. All data collection procedures prioritized the conservation and welfare of Vipera berus and aimed to contribute to the species’ long-term protection and understanding.ResultsDepth-variable line-upOf 140 positive targets, the dog correctly identified 127, missing 13 and producing one false alert at a negative target in seven runs (Supplementary Table 1). This results in an overall sensitivity of 90.71% and effectiveness of 86.39%, both well above chance levels (Supplementary Table 1). Sensitivity remained above 50% in all but two cases (Fig. 1). The GLM significantly outperformed the null model (Table 1; P < 0.001), showing the dog was more accurate with shed samples than swabs (P = 0.043). Sensitivity and effectiveness were higher for sheds (95.7% and 93.1%) than swabs (85.7% and 77.9%). Substrate type had no significant effect (P = 0.13).Table 1 Test statistics of the binomial GLM: response variable = search success/ correct indication (0/1), predictor = sample type, substrate type, depth [cm])Full size tableDepth had a significant impact on correct indications (Table 1, GLM, p = 0.00621). With increasing depth, sensitivity declined, while missed and false detections increased, especially in deeper depths (Fig. 1; Supplementary Table 1). Few false alerts occurred, even at 20 cm depth. Sensitivity remained above 80% in most tests but dropped sharply to 70% at 120 cm depth, while effectiveness decreased to 63.6%. (Supplementary Table 1).In peat–swab trials, sensitivity dropped from 100% to 40% at 100 cm depth, with three missed alerts recorded across five runs at 100 and 120 cm. In one of the three 100 cm trials with missed alerts, the dog gave no indication, while the remaining two included a false alert. Effectiveness also declined steeply beyond 100 cm, dropping to 28.6%, with a slight increase to 40% at 120 cm, representing a 71.4% reduction compared to 80 cm depth (Supplementary Table 2).Fig. 1The alternative text for this image may have been generated using AI.Full size imageChange in sensitivity [%] from 0 to 120 cm depth depending on treatment (a: stones/sheds; b: stones/swabs; c: peat/sheds; d: peat/swabs).Field testsOf the 70 positive samples, the dog correctly identified 69, including 35 on the pasture and 34 on the embankment. The single missed swab occurred due to a search that had to be terminated following an external disturbance. No false alerts were recorded, and blank samples were consistently ignored. Sensitivity and effectiveness reached 100% on the pasture and 97.1% on the embankment. Detection performance was slightly lower for fecal swabs (97.1%) than for sheds (100%), resulting in an overall sensitivity and effectiveness of 98.6%. The mean search time per positive sample was 37.5 s (SE = 3.18), with an average of 35.17 s on the pasture (median = 28 s) and 38.97 s on the embankment (median = 32 s) (Supplementary Fig. 2).Feasibility pilot testDuring pilot transect searches at the canal and in the Venner Moor, the dog accurately located the position of the previously tagged adders in their hibernacula, with indications within 50 cm of telemetry points. In the first canal search, the dog alerted twice (after 60 and 80 s) at a stone bar, with the second matching the hibernaculum location. In the Venner Moor, the dog indicated after 57 s on a peat moss bump, also matching the known hibernaculum site. In two control searches without tagged adders, the dog still indicated likely hibernaculum sites. The first alert (after 40 s) was at a stone bar; the second (after 130 s) was near a tree stump, where a potential burrow was later found.DiscussionOur results highlight the dog’s high detection accuracy and efficiency, especially in identifying shed samples, and underline the importance of considering depth when planning hibernacula surveys. Additionally, the dog’s success in both controlled and field searches suggests its potential for broader use in conservation efforts, particularly for species like adders in challenging environments, even though we are aware that general conclusions due to a single tested dog are limited.Depth-variable line-upThis study demonstrated that a specially trained detection dog could detect adder odor from both sheds and fecal swabs across various substrates in a depth-variable line-up. The high sensitivity and effectiveness (around 90%) achieved in these tests are comparable to other reptile line-up studies, such as those with tuatara, geckos, and green anole17,32. However, these studies are not fully comparable, as they did not use depth-adjustable line-ups, and the odor concentration remained constant in those studies, unlike the decreasing concentration in this study. A line-up study with increasingly diluted skin swabs from American bullfrogs found lower sensitivity and effectiveness, ranging from 49% to 87%25. Unlike the gecko study, the sensitivity for swabs was lower than for sheds, suggesting that fewer volatile organic compounds (VOCs) are released from swabs, thus reducing the detectable odor. This may be due to the larger sample size in the case of sheds, or the longer storage time of around four months, which could reduce the odor intensity, as observed in the American bullfrog study, where sensitivity dropped from 79% to 54% after four months25. Fresher swabs might yield better results, though it should be noted that the frog study used skin swabs, not fecal swabs. The relatively high standard error in this study also indicates some uncertainty in the results. The dog’s ability to detect adder odor under stones and peat at depths of up to 120 cm suggests that a trained dog could potentially locate adders in their hibernacula at the required depth and under typical substrate conditions7. However, detection performance significantly declined with increasing depth, a pattern that was also clearly visible in the course of sensitivity and effectiveness, particularly in the peat/swab tests. This decline in detection may be due to the difficulty of odor molecules diffusing through the substrate into the atmosphere, where they are detected by the dog. Although no significant effect of substrate type was detected in this study, it is reasonable to assume that VOC transport is facilitated in stone substrates via cracks and gaps, whereas in peat the movement of odor molecules through the soil matrix is slower and therefore likely requires more time for VOCs to reach the surface, particularly at very low concentrations. However, a real adder staying in the hibernacula for longer time than the odor samples tested in this study could potentially release more odor and smell more intense, making it potentially easier for the dog to detect even at higher depth. Factors like temperature, substrate affinity, and evaporation rates influence odor strength, while diffusivity is also affected by soil properties, moisture content, and bulk density33,34. As substrate thickness increases, the concentration of VOCs decreases, making detection harder. The sensitivity drops in peat and swab tests at depths over one meter, along with increased missed alerts, suggest, that insufficient VOC concentration passed through the peat to be detected. In contrast, sheds still had a higher detection rate at deeper depths, indicating that the VOC concentration was more detectable. False alerts may also occur due to residual odors from previous trials35. At greater depths, the VOC concentration of the positive sample might be comparable to residual odors, increasing the likelihood of false positives. In summary, the depth-variable line-up tests show that a dog can detect both odor types and highlight the influence of sample type and depth on detection. However, these laboratory findings may not fully translate to field conditions. The dog’s performance can be influenced by the test setup, and even in double-blind tests, handlers may unintentionally cue the dog by knowing that there is one positive sample24,36. Environmental factors such as temperature, wind speed, humidity, and terrain characteristics could also affect the dog’s performance in real-world searches19,20,21,37,38. Additionally, the PVC pipes used in the experiment may have optimised upward odor movement through a chimney effect38, thereby potentially enhancing odor availability compared to natural conditions. However, many adder hibernacula are naturally canal-like, such as vole burrows or gaps between stones, which could optimize odor dispersal for the dogs capability to detect it10. Also, dogs may differ individually for the maximum depth up to which they can detect odor, therefore a general conclusion regarding the mean maximum depth dogs can detect these samples needs to be tested with several dogs in future studies. As laboratory results may not always apply to field conditions, further outdoor testing is necessary17.Field searchesField searches achieved nearly 100% sensitivity and effectiveness, indicating reliable detection of both odor types at 50 cm depth under varying environmental conditions. Slightly better results than in line-up tests may reflect increased motivation due to the more engaging field setting, reducing errors linked to repetition and distraction39. Additionally, relict odors were avoided since transects were only used once. However, detection rates may decline with greater depth, depending on substrate, even though deeper hibernacula are unlikely in the embankment due to groundwater presence at 50–70 cm, as adders require dry hibernation sites7,10. Detection success and time were similar at both sites. While substrate effects remain poorly understood, results suggest detection is feasible in both clay loam and stone substrates. Previous studies report mixed outcomes: Human Remains Detection Dogs performed better in sandy soil, while a newt dog was more successful in clay, attributed to porosity differences and moisture content21,33. Peat, although more porous than loam clay, may impair diffusion due to high water content21,40,41. The effects of weather-related variables on detection performance could not be formally evaluated, as environmental conditions varied only within a narrow range during the experiments, precluding robust statistical modelling. Temperatures were generally moderate, which may have been beneficial not only for detection performance but also for dog welfare, as high temperatures can cause fatigue and increased panting, potentially impairing olfactory performance, while low temperatures may reduce odor release19,42. Similarly, the influence of wind speed on detection performance could not be statistically assessed due to insufficient variation in wind conditions. Although wind is known to affect odor dispersion, its potential effects may have been further reduced by the experimental design, including short transects, searches conducted on a long leash following a systematic search pattern, and odor retention in moist substrates, which can limit odor movement and reduce the relevance of wind effects19,43. No big difference was found in detection between sample types, unlike in line-up tests. The dog was only marginally slower on the embankment but exerted more effort navigating slippery stones, highlighting the importance of adjusting transect size to prevent fatigue, especially in older dogs17,39. Due to uneven terrain, the dog sometimes gave alternative alerts (e.g. sitting instead of lying down), stressing the need for handlers to correctly interpret subtle behavioral changes39,42,44. One missed alert occurred due to interference by an off-leash dog, underlining the importance of boldness and focus in detection dogs and public awareness about not disturbing working dogs39. The double-blind setup ensured reliable results, as the dog ignored all blank samples. This suggests detection was not influenced by human or carrier odors, a common issue in similar studies23,33,45. Future studies should aim to eliminate plastic tubes to avoid compound odor cues and mimic natural odor dispersal better (channel effect by tubes). Few comparable studies exist on subterranean animal odor detection. One involving great crested newts buried at 20 cm reached 88% success rate21. In another, dogs detected western indigo snake sheds and live animals both above and below ground, though odors dispersed through open burrows rather than closed substrate18,46. Our study extends this by confirming detection through largely closed substrate lacking any major openings. Our sensitivity exceeds the 75% threshold set for certified desert tortoise detection dogs43 confirming high perforance. Overall, results support that a dog can detect here tested reptile odors under mostly natural conditions with comparable accuracy to that for human remains or amphibians tested so far. In a future study, ideally, this should be tested without tubes and by testing several dogs. These findings have implications for locating other cryptic species such as smooth snakes. However, previous studies have shown that each species should be tested individually for its olfactory detectability by dogs, as some may have evolved chemical crypsis as an anti-predator and hunting strategy47.Pilot studySince odor samples represent only part of an animal’s scent, it was essential to test whether the adder detection dog could generalize this odor to living adders. The results of the pilot study suggest that such generalization is possible, as the dog successfully detected adders in natural hibernacula, indicating that trained odor cues remain detectable under natural conditions. However, of the two telemetrically confirmed adder locations, only the search in the Venner Moor can be considered an independent and valid alert. Although the dog also indicated the correct site at the Dortmund-Ems Canal during a double-blind search, the handler unintentionally directed the dog toward the location. This influence likely stemmed from prior survey knowledge and expectations held by the handler, which should be avoided in the future to reduce bias42. Dogs are highly responsive to subtle human cues due to their advanced social cognition33,48,49 and handler expectations can unintentionally lead to false alerts24. In contrast, during the Venner Moor search, neither the handler nor assistants had prior information. The dog searched independently and located the previously tagged adder’s hibernaculum without influence from human cues. The two alerts in transects without tagged adders suggest the likely presence of natural hibernacula. In the Venner Moor, one alert occurred near a tree stump on a peat dam. Nearby, a hole led into a burrow beneath the stump, consistent with known adder hibernaculum structures, such as mammal burrows and root systems7,10. While direct verification, as in the case of western indigo snakes, was not possible and is in general difficult for hibernacula and this type of search, the probability of a true positive was high. Detection of live adders in their hibernacula should be tested with several trained dogs and ideally with more tagged adders, if possible, to confirm the practical value of this non-invasive method, particularly as dogs may differ individually in their ability to generalize target odors50.Conclusions and consequences for future practical applicationOur study demonstrates, for the first time, that a detection dog can reliably identify adder scent in line-ups at depths of up to 120 cm and in the field at 50 cm under various substrates. To our knowledge, this is also the first successful detection of hibernating adders by a dog. Remarkably, the dog was trained solely with sheds and swabs and not live adders, which is a huge benefit for the animals’ welfare, especially when working with endangered species, in order to keep the disturbance to an absolute minimum during the training process. Although the handler–dog team was relatively inexperienced, potentially increasing the risk of unintentional handler bias during the pilot study, the searches were conducted under the guidance of an experienced wildlife detection dog handler and researcher to ensure methodological reliability. The results of the lineup study suggest that sheds were slightly easier to detect than swabs. When possible, sheds should be favored for training, as they are possible to collect non-invasively and safe without the need to handle a snake, particularly when working with rare or venomous species. However, since the dog was trained with both sample types, it remains unclear whether it would have been able to detect the animals after training with sheds only. This should be tested in the future. Cost-effectiveness plays a key role in the use of detection methods51. Wildlife detection dogs can be comparably priced to telemetry studies while being significantly less invasive, at least once the dog is trained52. Radiotelemetry only tracks a fraction of a population and is labor-intensive. Compared to visual adder surveys at basking sites, detection dogs can enhance both the quantity and quality of data collected, as they rely on scent rather than visibility53,54. This is particularly valuable in dense vegetation where visual surveys are limited37,55. A dog can also reduce spatial and gender-specific biases, as female adders emerge later in the season and are less frequently observed at their basking spots near the hibernacula7,10,56. While our study showed high sensitivity and efficacy rates for odor samples, this does not guarantee similar success with live adders. Previous studies, such as on indigo snakes, reported a 25% drop in detection performance when using live animals underground18. However, performance tends to improve with experience19. However, more tagged individuals are needed both during the training phase to generate greater experience in the dog, and in larger-scale studies to accurately assess detection rates during testing. As in other studies of cryptic species, verification remains challenging due to limited effectiveness of tools like endoscopes or wildlife cameras18,43. For practical use, guidelines from desert tortoise detection recommend a minimum 75% detection rate and multiple trained teams to validate results. Based on test performance, statistical models can estimate detection probabilities when one or more dogs indicate43. This approach could be adapted for adder surveys, ensuring dogs don’t follow each other’s scent. In summary, our findings highlight the potential of detection dogs to locate adder hibernacula and reduce mortality. They also show promise for broader application to other cryptic reptiles and subterranean habitats. However, expanded studies are needed to better quantify effectiveness and environmental influences.

Data availability

The datasets generated and/or analyzed during the current study are included in the supplementary materials accompanying this article.
ReferencesHochkirch, A. et al. A multi-taxon analysis of European Red Lists reveals major threats to biodiversity. PloS one. 18, e0293083. https://doi.org/10.1371/journal.pone.0293083 (2023).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Cox, N. et al. A global reptile assessment highlights shared conservation needs of tetrapods. Nature 605, 285–290. https://doi.org/10.1038/s41586-022-04664-7 (2022).Article 
ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Kelly, S. J. et al. Improving attitudes towards adders (Vipera berus) and nature connectedness in primary-age group children. People Nat. 5, 1908–1921. https://doi.org/10.1002/pan3.10545 (2023).Article 

Google Scholar 
Schmid, P. L., Grünewald, J., Förschler, M. I. & Griebeler, E. M. Substantial Genetic Differentiation Within and Between Populations of the European Adder (Vipera berus) in Baden-Württemberg. Ger. Ecol Evol. 15, e71644. https://doi.org/10.1002/ece3.71644 (2025).Article 

Google Scholar 
Emma Gardner, A. & Julian, C. M. & John Baker. Make the Adder Count: population trends from a citizen science survey of UK adders. Herpetol J, (2020).Ursenbacher, S., Monney, J. C. & Fumagalli, L. Limited genetic diversity and high differentiation among the remnant adder (Vipera berus) populations in the Swiss and French Jura Mountains. Conserv. Genet. 10, 303–315. https://doi.org/10.1007/s10592-008-9580-7 (2009).Article 

Google Scholar 
Otte, N., Bohle, D. & Thiesmeier, B. Die Kreuzotter. Ein Leben in ziemlich festen Bahnen: mit 105 Abbildungen, 29 Tabellen und 48 Farbtafeln 2nd edn (Laurenti, 2020).Podloucky, R., Koenzen, U. & Schlüpmann, M. in Rote Liste und Gesamtartenliste der Reptilien (Reptilia) Deutschlands, edited by Bundesamt für Naturschutz (Bundesamt für Naturschutz, pp. 44–45. (2020).Kollmann, J., Kirmer, A., Tischew, S., Hölzel, N. &amp; Kiehl, K. (2019). Renaturierungsökologie, Springer Spektrum.Schiemenz, H. Die Kreuzotter. Vipera berus 3rd edn (Westarp-Wiss;,, 1995). Spektrum Akad. Verl.Winkler, C., Pohlmann, P., Schmersow, I. & Krütgen, J. Renaturierung von Hochmooren in Schleswig-Holstein und Konflikte mit dem Reptilienschutz. RANA, 78–95 (2019).Völkl, W. Wasserstandserhöhungen bei der Moorrenaturierungen. Stellungnahme im Auftrag der Regierung von Schwaben Höhere Naturschutzbehörde, (2010).Cristescu, R. H. et al. Accuracy and efficiency of detection dogs: a powerful new tool for koala conservation and management. Sci. Rep. 5, 8349. https://doi.org/10.1038/srep08349 (2015).Article 
ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Grimm-Seyfarth, A., Harms, W. & Berger, A. Detection dogs in nature conservation: A database on their world‐wide deployment with a review on breeds used and their performance compared to other methods. Methods Ecol. Evol. 12, 568–579. https://doi.org/10.1111/2041-210X.13560 (2021).Article 

Google Scholar 
Cablk, M. E., Sagebiel, J. C., Heaton, J. S. & Valentin, C. Olfaction-based Detection Distance: A Quantitative Analysis of How Far Away Dogs Recognize Tortoise Odor and Follow It to Source. Sens. (Basel). 8, 2208–2222. https://doi.org/10.3390/s8042208 (2008).Article 
ADS 

Google Scholar 
Engeman, R. M., Vice, D. S., York, D. & Gruver, K. S. Sustained evaluation of the effectiveness of detector dogs for locating brown tree snakes in cargo outbound from Guam. Int. Biodeterior. Biodegrad. 49, 101–106. https://doi.org/10.1016/S0964-8305(01)00109-3 (2002).Article 

Google Scholar 
Fukuzawa, M. & Shibata, K. Testing the Ability of Dogs to Detect Different Odor Concentrations of the Carolina Anole (Anolis carolinensis) in Japan. Front. Vet. Sci. 7, 590834. https://doi.org/10.3389/fvets.2020.590834 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Stevenson, D. J. et al. Using a Wildlife Detector Dog for Locating Eastern Indigo Snakes (Drymarchon couperi). Herpetol Rev. 41, 437–442 (2010).
Google Scholar 
Grimm-Seyfarth, A. Environmental and training factors affect canine detection probabilities for terrestrial newt surveys. J. Vet. Behav. 57, 6–15. https://doi.org/10.1016/j.jveb.2022.07.013 (2022).Article 

Google Scholar 
Mutoro, N. M. et al. Impact of weather conditions on cheetah monitoring with scat detection dogs. J. Trop. Ecol. 37, 222–227. https://doi.org/10.1017/S0266467421000316 (2021).Article 

Google Scholar 
Glover, N. J., Wilson, L. E., Leedale, A. & Jehle, R. An experimental assessment of detection dog ability to locate great crested newts (Triturus cristatus) at distance and through soil. PloS one. 18, e0285084. https://doi.org/10.1371/journal.pone.0285084 (2023).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Lazarowski, L. et al. Effectiveness of marker training for detection dogs. Front. Vet. Sci. 12, 1538452. https://doi.org/10.3389/fvets.2025.1538452 (2025).Article 
PubMed 
PubMed Central 

Google Scholar 
Jezierski, T., Walczak, M. & Górecka, A. Information-seeking behaviour of sniffer dogs during match-to-sample training in the scent lineup. Pol. Psychol. Bull. 39, 71–80. https://doi.org/10.2478/v10059-008-0010-y (2008).Article 

Google Scholar 
Lit, L., Schweitzer, J. B. & Oberbauer, A. M. Handler beliefs affect scent detection dog outcomes. Anim. Cogn. 14, 387–394. https://doi.org/10.1007/s10071-010-0373-2 (2011).Article 
PubMed 
PubMed Central 

Google Scholar 
Matthew, E. E., Verster, R. & Weldon, C. A case study in canine detection of giant bullfrog scent. J. Vet. Behav. 69, 1–11. https://doi.org/10.25225/jvb.20043 (2020).Article 

Google Scholar 
Posit team. R Studio: Integrated Development Environment R (Posit Software, 2023).R Core Team. R: A language and environment for statistical computing (R Foundation for Statistical Computing).Hartig, F. CRAN: Contributed Packages (2016).Lüdecke, D., Ben-Shachar, M., Patil, I., Waggoner, P. & Makowski, D. performance: An R Package for Assessment, Comparison and Testing of Statistical Models. JOSS 6, 3139; (2021). https://doi.org/10.21105/joss.03139Wickham, H. et al. Welcome to the Tidyverse. JOSS 4, 1686. https://doi.org/10.21105/joss.01686 (2019).Article 
ADS 

Google Scholar 
Revelle, W. psych: Procedures for Psychological, Psychometric and Personality Research (Northwestern University, 2022).Browne, C. M., Stafford, K. J. & Fordham, R. A. The detection and identification of tuatara and gecko scents by dogs. J. Vet. Behav. 10, 496–503. https://doi.org/10.1016/j.jveb.2015.08.002 (2015).Article 

Google Scholar 
Alexander, M. B., Hodges, T. K., Wescott, D. J. & Aitkenhead-Peterson, J. A. The Effects of Soil Texture on the Ability of Human Remains Detection Dogs to Detect Buried Human Remains. J. Forensic Sci. 61, 649–655. https://doi.org/10.1111/1556-4029.13084 (2016).Article 
PubMed 

Google Scholar 
Fujikawa, T., Miyazaki, T., EFFECTS OF BULK DENSITY AND SOIL TYPE ON & THE GAS DIFFUSION COEFFICIENT IN REPACKED AND UNDISTURBED SOILS. Soil. Sci. 170, 892–901 ; https://doi.org/10.1097/01.ss.0000196771.53574.79 (2005).Article 
ADS 
CAS 

Google Scholar 
Alexander, M. B., Hodges, T. K., Bytheway, J. & Aitkenhead-Peterson, J. A. Application of soil in forensic science: residual odor and HRD dogs. Forensic Sci. Int. 249, 304–313. https://doi.org/10.1016/j.forsciint.2015.01.025 (2015).Article 
PubMed 

Google Scholar 
DeChant, M. T., Ford, C. & Hall, N. J. Effect of Handler Knowledge of the Detection Task on Canine Search Behavior and Performance. Front. Vet. Sci. 7, 250. https://doi.org/10.3389/fvets.2020.00250 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Paula, J. et al. Dogs as a tool to improve bird-strike mortality estimates at wind farms. J. Nat. Conserv. 19, 202–208. https://doi.org/10.1016/j.jnc.2011.01.002 (2011).Article 

Google Scholar 
Osterkamp, T. Detector Dogs and Scent Movement. How Weather, Terrain, and Vegetation Influence Search Strategies (CRC, 2020).Beebe, S. C., Howell, T. J. & Bennett, P. C. Using Scent Detection Dogs in Conservation Settings: A Review of Scientific Literature Regarding Their Selection. Front. Vet. Sci. 3, 96. https://doi.org/10.3389/fvets.2016.00096 (2016).Article 
PubMed 
PubMed Central 

Google Scholar 
ZAFFAR, M. & LU, S. G. Pore Size Distribution of Clayey Soils and Its Correlation with Soil Organic Matter. Pedosphere 25, 240–249. https://doi.org/10.1016/S1002-0160(15)60009-1 (2015).Article 

Google Scholar 
Rezanezhad, F. et al. Structure of peat soils and implications for water storage, flow and solute transport: A review update for geochemists. Chem. Geol. 429, 75–84. https://doi.org/10.1016/j.chemgeo.2016.03.010 (2016).Article 
ADS 
CAS 

Google Scholar 
DeMatteo, K. E., Davenport, B. & Wilson, L. E. Back to the basics with conservation detection dogs: fundamentals for success. Wildl Biol. wlb. 00584. https://doi.org/10.2981/wlb.00584 (2019).Harmon, R. S., Cablk, M., Nussear, K. & Esque, T. Validation and Development of a Certification Program for Using K9s to Survey Desert Tortoises; (2011). https://doi.org/10.21236/ADA551876La Jamieson, T. J., Baxter, G. S. & Murray, P. J. Who’s a Good Handler? Important Skills and Personality Profiles of Wildlife Detection Dog Handlers. Animals: open. access. J. MDPI. 8, 222. https://doi.org/10.3390/ani8120222 (2018).Article 

Google Scholar 
Edwards, T. L., Giezen, C. & Browne, C. M. Influences of indication response requirement and target prevalence on dogs’ performance in a scent-detection task. Appl. Anim. Behav. Sci. 253, 105657. https://doi.org/10.1016/j.applanim.2022.105657 (2022).Article 

Google Scholar 
Hyslop, N. L., Cooper, R. J. & Meyers, J. M. Seasonal Shifts in Shelter and Microhabitat Use of Drymarchon couperi (Eastern Indigo Snake) in Georgia. Copeia 2009, 458–464. https://doi.org/10.1643/CH-07-171 (2009).Article 

Google Scholar 
Miller, A. K., Maritz, B., McKay, S., Glaudas, X. & Alexander, G. J. An ambusher’s arsenal: chemical crypsis in the puff adder (Bitis arietans). Proc. R. Soc. B. 282, 20152182; (2015). https://doi.org/10.1098/rspb.2015.2182Hare, B. & Tomasello, M. Domestic dogs (Canis familiaris) use human and conspecific social cues to locate hidden food. J. Comp. Psychol. 113, 173–177. https://doi.org/10.1037/0735-7036.113.2.173 (1999).Article 

Google Scholar 
Miklósi, A. et al. A simple reason for a big difference: wolves do not look back at humans, but dogs do. Curr. Biol. 13, 763–766. https://doi.org/10.1016/s0960-9822(03)00263-x (2003).Article 
PubMed 

Google Scholar 
Caldicott, L. et al. Odour generalisation and detection dog training. Anim. Cogn. 27, 73. https://doi.org/10.1007/s10071-024-01907-0 (2024).Article 
PubMed 
PubMed Central 

Google Scholar 
Bennett, E. M., Hauser, C. E. & Moore, J. L. Evaluating conservation dogs in the search for rare species. Conserv. Biol. 34, 314–325. https://doi.org/10.1111/cobi.13431 (2020).Article 
PubMed 

Google Scholar 
Chambers, C. L., Vojta, C. D., Mering, E. D. & Davenport, B. Efficacy of scent-detection dogs for locating bat roosts in trees and snags. Wildl. Soc. Bull. 39, 780–787. https://doi.org/10.1002/wsb.598 (2015).Article 

Google Scholar 
Wasser, S. K. et al. Scat detection dogs in wildlife research and management: application to grizzly and black bears in the Yellowhead Ecosystem, Alberta, Canada. Can. J. Zool. 82, 475–492. https://doi.org/10.1139/z04-020 (2004).Article 

Google Scholar 
Arnesen, C. H., Johnsen, C. B., Costanzi, J. M. & Rosell, F. Canines (Canis lupus familiaris) as biodetectors for conservation work: Can they discriminate the rock ptarmigan (Lagopus muta) from the willow grouse (L. lagopus) in a yes/no task? PloS one. 15, e0228143. https://doi.org/10.1371/journal.pone.0228143 (2020).Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Sentilles, J., Vanpé, C. & Quenette, P. Y. Benefits of incorporating a scat-detection dog into wildlife monitoring: a case study of Pyrenean brown bear. J. Vertebr Biol. 69, 1–11. https://doi.org/10.25225/jvb.20096 (2021).Article 

Google Scholar 
Bohle, D. Aktivitätszeiträume von Reptilien im Havelland (Brandenburg) unter besonderer Berücksichtigung der Kreuzotter (Vipera berus). Z. f Feldherpetol. 23, 91–110 (2016).
Google Scholar 
Download referencesAcknowledgementsSpecial thanks go to Michael Schwartze for his support during the pilot study. We would also like to thank everyone who assisted us throughout this study. Our gratitude further extends to the many herpetological institutions that kindly provided sheds in advance of the experiments.FundingOpen Access funding enabled and organized by Projekt DEAL. The authors received no research funding.Author informationAuthors and AffiliationsInstitute of Landscape Ecology, University of Münster, Münster, GermanyMadita Schemel & Sascha BuchholzCentre for Integrative Biodiversity Research and Applied Ecology (CIBRA), University of Münster, Münster, GermanyMadita Schemel & Sascha BuchholzArtenspürhunde Schweiz, Olten, SwitzerlandJelena MausbachAuthorsMadita SchemelView author publicationsSearch author on:PubMed Google ScholarSascha BuchholzView author publicationsSearch author on:PubMed Google ScholarJelena MausbachView author publicationsSearch author on:PubMed Google ScholarContributionsM.S. and J.M. led the conception and design of the study, with significant contributions from S.B. J.M. also played a key role in structuring the study. M.S. was responsible for training the detection dog under J.M.‘s supervision. Line-up and field trials were conducted by M.S., with results validated by J.M. Data analysis and interpretation were carried out by M.S. with support from J.M. Statistical analyses were performed by M.S. in consultation with S.B. M.S. wrote the first draft of the manuscript, which was subsequently revised by S.B. and J.M. All authors contributed to drafting the manuscript and gave final approval for publication.Corresponding authorCorrespondence to
Madita Schemel.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Electronic Supplementary MaterialBelow is the link to the electronic supplementary material.Supplementary Material 1 (download DOCX )Supplementary Material 2 (download XLSX )Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and permissionsAbout this articleCite this articleSchemel, M., Buchholz, S. & Mausbach, J. Feasibility study on using a detection dog to localize hibernacula of Vipera berus.
Sci Rep 16, 14681 (2026). https://doi.org/10.1038/s41598-026-51778-3Download citationReceived: 10 November 2025Accepted: 29 April 2026Published: 09 May 2026Version of record: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51778-3Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
Keywordsreptilespopulation monitoringspecies conservationwildlife detection dogline-upscent More

AbstractRapid mid-winter and early spring warming events are emerging as a key but under-recognized driver of carbon loss from cold-region ecosystems. In mountain peatlands, their influence remains largely unknown. Here, we quantify the impact of chinook wind events—warm, dry downslope winds that can raise air temperatures by over 20 °C within hours—on ecosystem respiration in a montane peatland on the eastern slopes of the Canadian Rockies. Using eddy covariance carbon dioxide (CO2) flux measurements and meteorological data, 13 chinook events were identified between February and May 2021 and segmented each into pre-, during-, and post-event phases. A generalized additive mixed model accounting for temporal autocorrelation showed that CO2 emission rates increased significantly during chinook events and remained elevated afterward. CO2 emissions during snow-covered events in February-April were most likely driven by physical degassing from melting snow and thawing surface peat, whereas snow-free events in April and May likely reflected enhanced microbial activity in thawing peat. Our findings demonstrate that regularly occurring cold-season warming events can trigger substantial but short-lived CO2 releases from mountain peatlands, revealing a climate-sensitive carbon loss pathway likely to intensify as snowpack duration shortens and freeze-thaw regimes shift in mountain regions world-wide.

AcknowledgementsThis research was conducted on the traditional territories of the Blackfoot Confederacy, the Tsuut’ina Nation, and the Îyârhe Nakoda First Nations, and Métis Nation Region 3. We respectfully acknowledge and appreciate the interactions of living beings and land which have made these observations possible. We thank Adam Green for his invaluable assistance in processing the raw eddy covariance data, Lindsey Langs and Cob Staines for their help during fieldwork, Nichole-Lynn Stoll for the vegetation cover mapping, and Myroslava Khomik and Tyler Roman for their early feedback and guidance on data analysis. We thank Brian Menounos (Airborne Coastal Observatory, a Hakai Institute and University of Northern British Columbia program) who acquired and processed the drone imagery and LiDAR data.FundingThis work was funded by a Canada First Research Excellence Fund sub-grant (Mountain Water Futures), Canadian Foundation for Innovation grant (13163), Alberta Innovates Water Innovation Program grants (222301219 and 3360-E067), and an NSERC Discovery Grant (RGPIN-2017-05873).Author informationAuthors and AffiliationsDepartment of Geography and Planning, Global Institute for Water Security, University of Saskatchewan, Saskatoon, SK, S7N 5C8, CanadaMaría Elisa Sánchez & Cherie J. WestbrookDepartment of Geography & Environmental Management, University of Waterloo, Waterloo, ON, N2L 3G1, CanadaRichard M. PetroneAuthorsMaría Elisa SánchezView author publicationsSearch author on:PubMed Google ScholarRichard M. PetroneView author publicationsSearch author on:PubMed Google ScholarCherie J. WestbrookView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
María Elisa Sánchez.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary InformationBelow is the link to the electronic supplementary material.Supplementary Material 1 (download DOCX )Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleSánchez, M.E., Petrone, R.M. & Westbrook, C.J. Winter and early spring CO2 losses in a montane peatland are amplified by foehn winds.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-50725-6Download citationReceived: 04 October 2025Accepted: 23 April 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-50725-6Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsEddy covarianceCO2 fluxEcosystem respirationFreeze-thaw cyclesNon-growing seasonChinook winds More

AbstractAccurate global mangrove soil carbon estimates are essential, considering the critical role mangroves play in the coastal carbon cycle. However, current global estimates vary widely, ranging from 2.26 to 10.2 Pg C. Large uncertainties in carbon stocks can be challenging for effective policymaking and inaccurately estimate climate mitigation potential. Here, we identify factors driving spatial and landscape variability by comparing five global mangrove soil carbon stock models. Using the ensemble mean, we generate a global soil carbon stock estimate of 4.13 ± 0.89 Pg C (to 1 m depth). Significant (P < 0.001) spatial variability occurred in Africa and Asia, with high standard deviation between models and limited data representation. High standard deviation occurred in specific geomorphic settings, including terrigenous, deltaic, carbonate and open coasts, within microtidal sites and regions of low and high species diversity. Our findings can help guide future research and reduce uncertainties in carbon stock estimates.

Author informationAuthors and AffiliationsCoastal Studies Institute, East Carolina University, Wanchese, NC, 27981, USALucy Carruthers & David LagomasinoDepartment of Earth and Environmental Sciences, Tulane University, New Orleans, LA, 70118, USALukas Lamb-WottonSmithsonian Environmental Research Center, Edgewater, MD, 21037, USAJaxine WolfeDepartment of Earth, Environment, and Planning, East Carolina University, Greenville, NC, 27858, USAStuart E. HamiltonAuthorsLucy CarruthersView author publicationsSearch author on:PubMed Google ScholarLukas Lamb-WottonView author publicationsSearch author on:PubMed Google ScholarJaxine WolfeView author publicationsSearch author on:PubMed Google ScholarDavid LagomasinoView author publicationsSearch author on:PubMed Google ScholarStuart E. HamiltonView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
Lucy Carruthers.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary informationSupplementary information (download XLSX )Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleCarruthers, L., Lamb-Wotton, L., Wolfe, J. et al. Spatial and landscape-scale variability in global mangrove soil carbon estimates.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-51820-4Download citationReceived: 07 January 2026Accepted: 30 April 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51820-4Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative More

AbstractMicroplastics (MPs) are increasingly recognized as widespread contaminants in aquatic environments, including remote freshwater systems, where their presence is linked to waste generation and transport pathways. This study examined the occurrence, spatial–temporal distribution, polymer composition, and ecological risks of MPs in surface sediments from Black Lake and Devil Lake, two high-altitude glacial lakes in Durmitor National Park, Montenegro. Sediment samples were collected across three seasons and analyzed using standardized methods, including density separation, visual identification, and polymer characterization. MPs abundance averaged 5.1 ± 1.4 items per 100 g of dry sediment in Black Lake and 3.8 ± 0.5 items per 100 g in Devil Lake. Fibers and fragments were dominant morphotypes, with particles sized 1–3 mm prevailing. Blue particles were most frequent. Five polymer types were identified, with polyethylene as the dominant polymer. Pollution load index values indicated moderate contamination, while polymer hazard and ecological risk indices suggested high to very high environmental risk. The presence of MPs in protected, high-altitude glacial lakes highlights their vulnerability to diffuse pollution sources, including tourism and long-range transport. The findings provide baseline data for alpine freshwater environments and underline the importance of integrating MPs pollution into waste management and environmental protection strategies.

AcknowledgementsThe authors acknowledge the research core funding No. I0-0003 Infrastrukturna dejavnost KI.Author informationAuthors and AffiliationsNational Institute of Chemistry, Hajdrihova 19, Ljubljana, 1000, SloveniaNeda Bošković, Ivan Jerman & Andrej RaceFaculty of Technology, University of Montenegro, Cetinjski put b.b, Podgorica, 81000, MontenegroŽeljko JaćimovićFaculty of Health Sciences, University of Ljubljana, Zdravstvena Pot 5, Ljubljana, 1000, SloveniaPolonca TrebšeMarine Biology Station, National Institute of Biology, Fornače 41, Piran, 6330, SloveniaOliver BajtFaculty of maritime studies and Transport, University of Ljubljana, Pot pomorscakov 4, Portoroz, 6320, SloveniaOliver BajtAuthorsNeda BoškovićView author publicationsSearch author on:PubMed Google ScholarŽeljko JaćimovićView author publicationsSearch author on:PubMed Google ScholarIvan JermanView author publicationsSearch author on:PubMed Google ScholarAndrej RaceView author publicationsSearch author on:PubMed Google ScholarPolonca TrebšeView author publicationsSearch author on:PubMed Google ScholarOliver BajtView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
Neda Bošković.Ethics declarations

Competing interests
The authors declare no competing interests.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleBošković, N., Jaćimović, Ž., Jerman, I. et al. Microplastic pollution and ecotoxicological risk in high-altitude glacial lakes of Durmitor, Montenegro.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-51455-5Download citationReceived: 19 January 2026Accepted: 28 April 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51455-5Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsMicroplasticGlacial lakesSediment pollutionDurmitor National ParkEcological risk assessment More

AbstractClimate variability poses serious threats to agricultural productivity and food security, particularly for smallholder rice farmers in climate-vulnerable regions such as northwest Bangladesh. This study examines the adoption of multiple climate change adaptation strategies among smallholder rice farmers in this region. Using a multistage sampling technique, data were collected from 498 farmers. A multivariate probit model and a fractional response model were applied to analyze both the determinants and intensity of adoption of key strategies, including adjusting planting dates, integrated pest management, tolerant rice varieties, improved rice varieties, and rice straw mulching. This joint modeling framework captures interdependence among strategies and provides deeper insights than single-strategy approaches. Robustness was checked using a multinomial logistic regression model. The results indicate that socio-economic factors such as farming experience, access to credit, and membership in farmer groups significantly influence adoption behavior. Strong complementarities among adaptation strategies are also observed, suggesting that farmers tend to adopt these practices in bundles rather than individually. The findings contribute to the literature by highlighting the effectiveness of bundled adaptation strategies in enhancing resilience. They offer important policy implications for designing integrated and context-specific interventions to strengthen adaptive capacity and promote resilient agricultural systems in climate-exposed regions.

AcknowledgementsThe authors are grateful to the rice farmers of northwestern Bangladesh for their valuable cooperation and continuous support throughout this research. Their willingness to participate and share their experiences made this study possible.Author informationAuthors and AffiliationsDepartment of Economics, Pabna University of Science and Technology, Pabna, 6600, BangladeshMd Shohidul Islam, Bikash Chandra Ghosh & Apurba AdhikaryAuthorsMd Shohidul IslamView author publicationsSearch author on:PubMed Google ScholarBikash Chandra GhoshView author publicationsSearch author on:PubMed Google ScholarApurba AdhikaryView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to
Bikash Chandra Ghosh.Ethics declarations

Competing interests
The authors declare no competing interests.

Ethics Approval
Ethical approval to conduct this study was obtained from the Research Ethics Committee of the Department of Economics, Pabna University of Science and Technology, Rajapur, Pabna-6600, Bangladesh (Approval No. PUST/ECON/REC/136). All methods were carried out in accordance with the ethical guidelines for social science research at Pabna University of Science and Technology.

Consent to Participate

Informed verbal consent was obtained from all participants prior to the interviews. Participation was voluntary, and respondents were assured of anonymity and confidentiality.

Consent to Publish
Participants consented to the publication of anonymized and aggregated data for academic purposes.

Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleIslam, M.S., Ghosh, B.C. & Adhikary, A. Interdependent adoption of climate change adaptation strategies among rice farmers in northwest Bangladesh.
Sci Rep (2026). https://doi.org/10.1038/s41598-026-51096-8Download citationReceived: 10 January 2026Accepted: 25 April 2026Published: 09 May 2026DOI: https://doi.org/10.1038/s41598-026-51096-8Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy shareable link to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsClimate changeAdoption strategiesMultivariate probit modelFractional response modelNorthwestern Bangladesh More