Stable isotope tracking

Stable isotope analysis is a well-established tool in (paleo-)ecological research, used to infer trophic relationships and dietary patterns over time (e.g., [21,28,42–44]). By analysing the ratios of naturally occurring stable isotopes, most commonly carbon and nitrogen, and sulfur, in biological tissues, it is possible to reconstruct feeding ecology over extended periods, as, e.g., hair integrate isotopic signals during its growth [44]. Carbon isotopes (δ¹³C) provide information on the primary sources of dietary carbon and are useful for distinguishing between prey from different habitats or consuming plants using different photosynthetic pathways (e.g., forest vs. open-field species, C₃ vs. C₄ plants) [40,42,45,46]. Nitrogen isotopes (δ¹⁵N) reflect an organism’s trophic position, with higher δ¹⁵N values indicating higher positions in the food web [40,47]. Sulfur isotopes (δ³⁴S) can help differentiate between terrestrial and aquatic food sources, or can be used as regional tracking isotope [40,48,49]. It is important to note that these isotopic signals are primarily incorporated from the protein fraction of the diet; therefore, stable isotope analysis from hair allows for the reconstruction of protein sources rather than the complete dietary composition. Interpreting isotope values in carnivores, however, is more challenging than in herbivores, as carnivores typically consume a variety of prey species that may come from different habitats and trophic levels. While herbivore isotope signatures are closely tied to local vegetation and thus reflect habitat use more directly, carnivore isotopic profiles represent an averaged signal of their prey’s isotopic values. This complexity necessitates cautious interpretation; however, stable isotope analysis can still reveal meaningful ecological patterns, such as preferred hunting grounds or dietary specialization and overlap [50,51].

In the case of European wildcats, stable isotope analysis can offer valuable insights into both habitat use and dietary flexibility (Fig 2). δ¹³C values may indicate whether individuals predominantly hunt in forested areas, open landscapes, or agricultural zones, based on the habitat affiliation of their prey. δ¹⁵N values can reveal trophic shifts, such as a reliance on rodents and small vertebrates versus the inclusion of anthropogenic food sources. Elevated δ¹⁵N values may arise either from prey that themselves feed on human-derived resources (e.g., commensal rodents in agricultural or urban environments) or from the direct consumption of anthropogenic material such as food waste or pet food. Finally, δ³⁴S values may help detect the consumption of aquatic or marine-derived resources; for instance, the presence of marine sulfur signatures in inland wildcats would strongly suggest access to human-related food sources such as food waste or pet food.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Schematic representation of potential trophic strategies of the European wildcat.

A) Primary prey (small mammals) derived from forest habitats, B) mixed diet including small mammals from both forest and open land habitats, C) prey mainly derived from open land habitats. Values in brackets indicate the typical δ¹³C range of deep forest and meadow/agriculture areas.

https://doi.org/10.1371/journal.pone.0343705.g002

Aim and case studies

This study aims to evaluate the potential of retrospective stable isotope monitoring to investigate anthropogenic influences on the trophic ecology of European wildcats. By integrating two complementary case studies, we explore how human-induced factors, such as hybridization, landscape modification, and agricultural practices, shape wildcat foraging behaviour over space and time. Our overarching goal is to demonstrate how stable isotope data from hair can be used to reconstruct dietary patterns, assess niche dynamics, and identify drivers of ecological change in carnivores.

The first case study is based on hair samples (n = 31) collected in spring 2021 using the lure stick method in two regions of Germany (Fig 1A & B): the Markgräflerland in Baden-Württemberg and the Taunus region in Hesse. These samples were originally collected in the frame of the national wildcat monitoring program. The two populations differ markedly in their conservation histories: the Taunus population represents a stable, long-established wildcat population with a low hybridization rate (~2%), whereas the Markgräflerland population is the result of recent natural recolonization from neighbouring French populations and exhibits a much higher hybridization rate (~43%) [30]. Against this background, we investigated how wildcats, hybrids, and domestic cats partition trophic space within each region, and to what extent hybrids and domestic cats overlap with wildcats under contrasting hybridisation regimes, thereby evaluating hybridization as a potential anthropogenic driver of trophic change.

The second case study focuses on 47 hair samples from deceased wildcats and wildcat-like domestic cats collected between 1995 and 2021 across various regions of Thuringia (Fig 1C), sourced from the Phyletic Museum in Jena. These regions encompass a broad gradient of landscape types, ranging from predominantly forested and structurally complex landscapes (Thuringian Forest, Hainich), to agriculturally intensive lowland areas with a high proportion of arable land and comparatively low forest cover (Thuringian Basin, Harz Foreland), as well as more heterogeneous forest–field mosaics characterized by mixed land use, smaller field sizes, and a higher density of edges and ecotones (East Thuringia). Through stable isotope analysis of carbon, nitrogen, and sulfur, we reconstructed temporal dietary trends and investigated how wildcat foraging behaviour may have shifted over time in response to regional climate change, agricultural development, and other landscape-level changes. We also assessed relationships between trophic variation and individual traits such as body size and sex.

Together, these case studies provide a framework to assess the trophic responses of wildcats to multiple dimensions of anthropogenic pressure.

Trophic ecology of wildcats, domestic cats and their hybrids

While based on a somewhat limited set of hair samples, our isotopic results provide nuanced insights into the trophic ecology of European wildcats, domestic cats, and their hybrids. While wildcats generally exhibited the narrowest isotopic niche, suggesting a more specialized diet largely reliant on forest-dwelling small mammals (low δ¹³C and δ¹⁵N values [54,55]), this pattern was not uniform across space and time. For instance, in the Markgräflerland, wildcats displayed a highly constrained isotopic niche, indicative of strong ecological specialization based on hair samples collected in spring 2021. However, this constrained niche likely reflects a temporally limited pattern of resource use rather than long-term ecological specialization. As the Markgräflerland population represents a recently recolonized range-edge population, the narrow isotopic niche may capture short-term foraging behaviour under locally abundant and predictable prey conditions during a single season, rather than reduced trophic flexibility. In contrast, long-term data from Thuringian wildcats revealed a temporal increase in δ¹³C values, suggesting a gradual dietary shift from forest-dwelling prey toward species associated with agricultural habitats, such as crop-feeding rodents. This trend, together with the large spatial extent and environmental heterogeneity of habitats within Thuringia, likely resulted in the expanded δ¹³C range among wildcats observed here. Nevertheless, this range remained smaller than those observed in hybrids or domestic cats, whose diets are generally broader and more strongly shaped by anthropogenic food sources. This progressive niche expansion mirrors findings from Italy [56], where wildcats increasingly consumed prey associated with human-altered landscapes, and is supported by studies highlighting the species’ adaptability to fragmented or agriculturally influenced habitats [30,57].

Our data on the Thuringian deceased cats also reveal distinct sex- and size-related patterns in wildcat trophic ecology. Females exhibited significantly higher δ¹³C values compared to males, and δ¹³C values correlated negatively with body weight, total length, and hind-foot length. This indicates that smaller individuals, particularly females, tend to forage more frequently in open or edge habitats, where prey species associated with agricultural or mosaic landscapes are more abundant. Oliveira et al. [58] demonstrated that female wildcats show a stronger selection for high-quality habitats with greater structural heterogeneity and prey availability, such as scrubland–agriculture mosaics. Similarly, Beugin et al. [59] found that females predominantly occupy forest interiors with direct access to field edges, allowing optimal foraging–shelter dynamics. In contrast, males exhibited consistently lower δ¹³C values, which we interpret as a stronger reliance on forest-dwelling prey species and a preference for structurally closed habitats. The negative correlation between δ¹³C and body size supports this view, suggesting that larger individuals, mainly males, forage predominantly within forest interiors characterized by stable, C₃-based food webs and limited anthropogenic influence. This interpretation aligns with spatial ecology studies indicating that males range widely across forest-dominated territories [58,59]. Additionally, δ¹⁵N values were positively correlated with body size, implying that larger individuals, predominantly males, may partially feed on prey from higher trophic levels. This may reflect a shift toward larger prey items or small omnivores and insectivores, including birds, in larger individuals, while smaller wildcats appear to rely more heavily on voles and other low-trophic rodents. These patterns align with general carnivore ecology and the sex-specific foraging strategies commonly observed in solitary predators [60].

Anthropogenic influence on wildcat’s trophic behaviour

Anthropogenic impact, particularly through agriculture and hybridization, emerges as an important driver of trophic dynamics in European wildcats. Hybrids showed substantial isotopic niche overlap with wildcats overall (72.8%), reaching 93% in the Markgräflerland and 68.3% in the Taunus. This pronounced overlap, especially in the Taunus, where genetic studies report only low hybridisation rates (~2%, [7]), indicates that hybrids frequently exploit similar trophic resources as wildcats and cannot be considered trophically distinct. In contrast, domestic cats exhibited minimal overlap, averaging only 5.9%, with 0% overlap in the Markgräflerland and similarly low overlap values observed in the Thuringian reference dataset (2.6%), suggesting limited trophic similarity and competitive pressure. These results align with findings from Germain et al. [61], who reported dietary compositions of hybrids in northeastern France as intermediate between wildcats and domestic cats, with considerable niche overlap toward wildcats. Similar conclusions were reached by Biró et al. [33] using traditional dietary analyses of genetically identified wildcats, hybrids, and feral domestic cats, providing independent support for substantial trophic overlap between wildcats and hybrids. Similarly, Tiesmeyer et al. [9] confirmed regionally variable hybridisation across Europe. Their study highlights the need for regional assessments of ecological consequences, such as niche displacement or functional redundancy. In the Markgräflerland, where hybridisation is frequent (~43%, [13]), hybrids not only overlapped substantially with wildcats but also occupied the largest isotopic niche area, reflecting greater dietary flexibility and ecological integration. In our dataset, hybrids from the Taunus were predominantly backcrosses to wildcats, but also included one F₁ individual, whereas hybrids from the Markgräflerland consisted of F₁ and F₂ generations. These differences in hybrid composition may also help to explain the varying degree of trophic overlap observed between regions. In the Taunus, where backcrosses to wildcats dominate, the isotopic niche of hybrids closely mirrors that of wildcats, reflecting similar foraging strategies and resource use. In contrast, in the Markgräflerland, where earlier-generation hybrids and backcrosses to domestic cats occur, hybrids displayed not only the highest overlap with wildcats but also the largest isotopic niche area. This broader niche likely reflects the combined trophic flexibility inherited from both parental lineages and a higher propensity to exploit anthropogenic food sources. Such differences underline that the ecological consequences of hybridisation depend not only on its frequency but also on the generational composition of hybrid cohorts. These patterns underscore the link between hybridisation, landscape modification, and shifts in ecological roles, particularly as wildcats expand into human-altered environments and exhibit more synanthropic behaviour [62].

To further investigate this synanthropic relationship, we examined whether regional agricultural land use could help explain the observed δ¹³C trends and isotopic niche dynamics in Thuringian wildcats. In our long-term Thuringian dataset, wildcats showed a gradual increase in δ¹³C values over time, indicating a growing dietary contribution from prey associated with agricultural environments, particularly rapeseed, pasture, and cereal cultivation. Agricultural areas play a dual role in wildcat ecology: on the one hand, they structurally shape the landscape and may provide attractive edge habitats; on the other hand, they compete with structured, undisturbed forests, which are traditionally considered the preferred habitat of European wildcats [3,63]. However, the isotopic niches observed in our data suggest a degree of ecological plasticity, with individuals, particularly during summer, exploiting food resources in open areas with increasing intensity over the years. Our isotopic findings are consistent with recent spatial ecological studies suggesting that European wildcats are increasingly utilizing agriculturally dominated landscapes [10,30,57]. Ruiz-Villar et al. [57] demonstrated that wildcat home range size increases in regions with a high proportion of intensive agriculture and low forest edge density, highlighting the need for larger foraging areas due to reduced prey availability in rather monotonous landscapes. Notably, the presence of heterogeneous features like forest edges buffered the negative effects of agricultural intensification, suggesting that structurally complex agricultural mosaics can still support wildcat populations. Our data align with habitat use patterns described by Jerosch et al.[64], who found that wildcats, particularly females, rely heavily on small-scale shelter structures such as hedgerows, fallow fields, and ecotones to navigate and persist within open landscapes. Their results also indicated seasonal shifts in habitat use, with greater tolerance for open habitats during summer when crops provide temporary cover. Our seasonal isotope data similarly show that δ¹³C values were higher in summer than in winter, suggesting increased foraging in open agricultural areas during the growing season. In contrast to the clear patterns observed for δ¹³C, the weak correlations observed for δ³⁴S across environmental and individual variables further support a cautious interpretation of sulfur isotope data in this study. While δ³⁴S has the potential to indicate aquatic or marine-derived food sources or anthropogenic inputs, such signals were not consistently expressed in our dataset. This likely reflects the dominance of terrestrial prey and regionally stable sulfur baselines in the studied systems, rather than an absence of anthropogenic influence per se.

Taken together, these findings strongly indicate adaptive trophic responses of wildcats to anthropogenic landscape change. While such plasticity may facilitate persistence in human-dominated habitats, it may also increases the potential for hybridisation with domestic cats and increasing ecological overlap hybrids, posing long-term challenges for the conservation of the wildcat’s ecological and genetic integrity.

Retrospective isotopic monitoring as a tool for carnivore trophic ecology

The combined insights from our two case studies, spanning both spatial contrasts and temporal dynamics, highlight the value of retrospective stable isotope analyses for understanding carnivore trophic ecology in complex and changing landscapes. By integrating isotopic data across taxa, time periods, and environmental gradients, this approach enables the reconstruction of dietary patterns, niche shifts, and individual-level variation that may otherwise remain undetected. Our findings demonstrate how stable isotope data derived from archived or opportunistically collected material can reveal not only ecological differentiation among wildcats, domestic cats, and hybrids, but also track subtle responses of wildcat trophic behaviour to anthropogenic pressures such as hybridisation and agricultural intensification. These applications underscore the potential of retrospective isotopic monitoring as a cost-effective, scalable, and non-invasive tool for advancing carnivore ecology and informing long-term conservation strategies. Retrospective isotopic monitoring enables the reconstruction of ecological responses over time, particularly when direct behavioural or dietary data are unavailable. By analysing stable isotope signatures in archived tissues, such as hair, bone, or teeth, researchers can infer past trophic relationships and habitat use with fine temporal resolution. This approach has already proven useful in a variety of systems: for instance, Turner et al. [65] demonstrated how retrospective isotope data revealed ecosystem responses to hydrological regulation over multiple decades in a riverine food web. Similarly, studies in boreal forests showed that δ¹³C and δ¹⁵N values in mammal tissues can reliably track shifts in foraging behaviour following habitat disturbance, underscoring the method’s sensitivity to land-use change [66–69]. Our findings confirm that stable isotope signatures in wildcat hair reflect both ecological specialization and increasing anthropogenic integration. Importantly, by combining spatial and temporal datasets, we were able to detect subtle shifts in δ¹³C values that would likely be missed in short-term studies. Moreover, the inclusion of δ³⁴S adds another layer of resolution by potentially capturing baseline shifts due to atmospheric deposition or fertilizer use, which further enhances the method’s applicability in agricultural landscapes [70]. This aligns with observations by Crawford et al. [69], who emphasized that stable isotope approaches can provide long-term baselines for mammalian ecology and highlight delayed responses to environmental pressures. In this context, isotopic monitoring is particularly relevant for protected and elusive carnivores like the European wildcat, whose behavioural adaptations and ecological roles are increasingly shaped by human activity.

Limitations and future directions

Despite the valuable insights gained, some limitations of this study should be acknowledged. Sample sizes within the individual taxa and regions were relatively small, particularly for hybrids, which limits the statistical resolution and the detection of subtle trophic or seasonal effects. In addition, uneven sample sizes among regions may influence range-based niche metrics such as total area (TA) and isotopic ranges, which should therefore be interpreted cautiously; accordingly, greater emphasis was placed on core niche (SEAc) and nearest-neighbour metrics that are less sensitive to sample size effects. Moreover, the degree and direction of hybridisation (F₁, F₂, or backcrosses to either wildcats or domestic cats) could not be included in the statistical analyses, as this information was linked to different areas (Taunus vs. Markgräflerland). The dataset also comprised a heterogeneous distribution of sexes and seasons, which may have influenced isotopic variation and limited the comparability among groups. Nevertheless, the results provide a robust first overview of the potential of stable isotope analyses on hair as a tool for nature conservation and ecological monitoring. Future studies should aim for broader seasonal coverage and more balanced sampling designs, as marked seasonal differences in foraging behaviour are likely among wildcats, domestic cats, and hybrids. In addition, the analysis of multiple sections of continuously growing tissues, such as tactile hairs (vibrissae) or sequential hair segments, could provide higher temporal resolution and allow the reconstruction of short-term dietary shifts within individuals. Integrating isotopic, genetic, and spatial data will further enhance our understanding of how hybridisation and anthropogenic landscape change jointly shape the trophic ecology of the European wildcat.

Elemental and isotopic analyses

Elemental and isotopic measurements were carried out at the Geoecology Stable Isotope Platform at the University of Tübingen, using a Vario Isotope Cube elemental analyzer in conjunction with an IsoPrime Vision isotope ratio mass spectrometer (IRMS). We used the international references V-PDB for carbon, atmospheric nitrogen (AIR) for nitrogen, and V-CDT for sulfur isotope ratios to calibrate the measured samples. The international laboratory standards USGS-40 (δ¹³C = −26.39 ‰; δ¹⁵N = −4.52‰) and USGS-41a (δ¹³C = +36.55‰; δ¹⁵N = +47.55‰) on one hand and IAEA-S1 (δ³⁴S = −0.30‰), IAEA-S2 (δ³⁴S = +22.62‰) and IAEA-S3 (δ³⁴S = −32.49‰) on the other hand, as well as two in-house reference materials (modern collagen of camel: δ¹³C = −14.8‰; δ¹⁵N = +8.1‰, δ³⁴S = +13.63‰, and elk: δ¹³C = −23.9‰; δ¹⁵N = +2.6‰, δ³⁴S = +6.68‰) were used to track device stability, and do drift correction. An analytical error below 0.1‰, 0.2 ‰, and 0.4‰ respectively (1σ) was determined for δ¹³C, δ¹⁵N, and δ³⁴S in all the repeated analyses. The reproducibility error for the amounts of C and N was lower than 1%, and lower than 2% for S.

Atomic elemental ratios were calculated using the following formula:

Atomic C/N ratios between 2.9 and 3.6 are accepted [72], while for atomic C/S ratios values between 300 and 900, and for atomic N/S ratios values between 100 and 300 are tolerated [73].

The isotopic ratios are expressed using the δ (delta) value as follows:

We applied a lipid correction to all measured δ¹³C values to exclude the isotopic signature of lipids in the hairs. Because lipids are typically depleted in ¹³C relative to proteins, uncorrected values can bias carbon isotope ratios toward lower δ¹³C values [74]. Since no chemical lipid extraction could be performed prior to isotope measurement, δ¹³C values were lipid-corrected post hoc using an established equation [75].

This correction accounts for the relationship between lipid content and isotopic composition in mammal hair, ensuring comparability among samples with varying lipid levels. Furthermore, we applied a Suess effect correction to all measured values, based on global atmospheric δ¹³C trends (see Scripps CO₂ Program [76]), using pre-industrial conditions as the baseline to ensure comparability among samples [77]. The Suess effect refers to the decline in the atmospheric δ¹³C signature caused by large-scale combustion of ¹³C-depleted fossil fuels. Since the onset of the industrial revolution, this process has progressively lowered δ¹³C values in atmospheric CO₂ and, consequently, in primary producers and higher trophic levels. Correcting for this effect is essential in stable isotope studies involving modern comparative data, as it enables direct comparison with pre-industrial and archaeological/paleontological samples. Without such a correction, temporal shifts in δ¹³C could be misinterpreted as ecological or dietary changes rather than reflecting anthropogenic alterations of the global carbon cycle.

Statistical analysis

All statistical analyses were performed in R (version 4.5.1; R Core Team) using RStudio (version 2025.05.1 Build 513; Posit Software). The full R script is available on the open online data repository zenodo.org (https://doi.org/10.5281/zenodo.17589796). Graphical preparation of figures was conducted with Affinity Designer 2 (version 2.6; Serif Europe Ltd.).

Stable isotope data (δ¹³C, δ¹⁵N, δ³⁴S) were first quality-checked by excluding samples with atomic C/N ratios outside the range 3.0–4.05, %C values > 50, %N values > 20, or missing regional assignments [44,72]. Cleaned data were used to generate descriptive statistics (mean, SD, min, max) for each taxon and region. Differences in isotopic values between taxa (wildcats, domestic cats, and hybrids) were tested using pairwise Wilcoxon rank-sum tests with Benjamini–Hochberg correction for multiple comparisons. To quantify isotopic niches, we applied the Stable Isotope Bayesian Ellipses in R (SIBER [53]) framework. Bayesian standard ellipse areas (SEAb) were estimated using Markov Chain Monte Carlo simulations (20,000 iterations, 3 chains, burn-in = 1,000, thinning = 10). Maximum-likelihood ellipse metrics were calculated for each group, and niche overlap was quantified using maximum-likelihood overlap functions. In addition, Layman metrics (δ¹³C and δ¹⁵N ranges, total area (TA), centroid distance (CD), mean nearest neighbour distance (NND), SDNND) were calculated to assess trophic niche width and structure [52]. We calculated both the corrected standard ellipse area (SEAc), which represents approximately 40% of the data and describes the core isotopic niche, and the TA, which encompasses all data points including outliers and thus reflects the full extent of niche width [53]. Linear models and robust linear models (rlm) were fitted to estimate temporal slopes of δ¹³C, δ¹⁵N, and δ³⁴S. To assess seasonal differences in isotopic signatures, hair samples were assigned to summer-grown or winter-grown growth periods based on the month of collection. European wildcats and domestic cats undergo two main moulting phases per year, in spring and autumn, and hair isotope values integrate dietary information over several months during growth. Following this biology, samples collected between December and April were classified as summer-grown hair, reflecting growth during the preceding spring–summer period, whereas samples collected between May and November were classified as winter-grown hair, reflecting growth during autumn and winter. To test for sex-specific and body-size-related differences, two-sample t-tests and regression models were performed. Seasonal differences (summer vs. winter hair growth) were assessed using stratified regression analyses.

Meteorological data (monthly mean temperature, precipitation) were obtained from the German Weather Service (DWD) for 1994–2022 and aggregated at the subregion level. Seasonal subsets (summer: June–August; winter: December–February) were analysed separately. Linear regression models were fitted per subregion to quantify trends. Agricultural land-use data (cereal, maize, rapeseed, pasture area) were extracted from the Thuringian State Office for Statistics. Finally, relationships between isotopic values (δ¹³C, δ¹⁵N, δ³⁴S) and explanatory variables (climate, agricultural land use, sex, body size, age, latitude, longitude) were explored using Pearson correlation analyses and simple linear regression models. Correlation analyses were restricted to wildcats from the Thuringian dataset and were based on pairwise complete observations. Sex was coded as a numeric variable (males = 2, females = 1). No multivariate or mixed-effects models were applied due to limited sample sizes and collinearity among explanatory variables.