Fig 1.
Study areas in the Iberian Peninsula where wild carnivore scats were collected to evaluate the reliability of scat identifications.
Black polygons show the limits of the five protected areas surveyed: Aracena, Cazorla, Montaña Palentina, Sierra Nevada, and Ordesa. The background map shows the main climatic categories of the Iberian Peninsula according to the Köppen–Geiger classification (obtained from [11]), illustrating the climatic context in which the study areas are located. A detailed description of the climatic categories affecting each study area is provided in the Supporting Information.
Table 1.
Fixed‐effect estimates from the binomial GLMM predicting the probability of genetic identification of scats in seven carnivore species from the Iberian Peninsula. Odds ratios (OR), 95% Wald confidence intervals (CI), and p‑values are shown. The reference levels are Scat Age = Medium, Observer = Observer 1, and Year = 2022. Significant predictors (p < 0.05) include fresh scat age, precipitation, mean temperature, Observer 3, Observer 4, Observer 5, and Year 2023.
Fig 2.
Combined panel of marginal effects: (a) Scat age, (b) Precipitation, (c) Temperature, and (d) Year. Each panel shows predicted probabilities with 95% CIs, derived from the GLMM while integrating over remaining predictors.
Fig 3.
Predicted probabilities (±95% CI) for each observer (Observer_2, Observer_3, Observer_4, Observer_5) on genetic success.
Estimates are adjusted for all other covariates.
Fig 4.
Heatmap of concordance between field-based and genetic species identifications.
Each cell shows the count of scats assigned to a given species in the field (rows) and by genetic analysis (columns). Diagonal cells (shaded in darker blue) indicate matches between field and genetic identifications. Off-diagonal cells reveal misclassifications or broader-level genetic identifications (e.g., Felis sp. for field-identified F. silvestris). Color intensity corresponds to the number of samples in each cell.
Table 2.
Concordance rates between field-based and genetic identifications for Martes foina and Martes martes scats in different scenarios.
Table 3.
Odds ratios (OR) and 95% Wald confidence intervals (CI) for predictors of correct field identification of scats. Estimates are derived from a generalized linear mixed-effects model (logit link) with area as a random intercept. The reference levels are Observer_1, Fresh scat age, High certainty, and Win_22 sampling session. P-values are from Wald z-tests. The effect of the area was minimal as σ²Area = 0.034 (σ = 0.185), adjusted ICC = 0.01.
Fig 5.
Predicted probability of correct field identification across key categorical predictors.
Panels show marginal effects (point estimates ± 95% confidence intervals) from the final GLMM with area as a random effect. (a) Field‐identification certainty. (b) Observer identity. (c) Sampling session. (d) Scat age. All probabilities are back‐transformed on the response scale from the logit link. Error bars indicate 95% Wald confidence intervals.