Beyond habitat suitability: integrating ecological stress and predictive uncertainty to prioritize conservation areas for threatened Atlantic Forest trees

This dataset contains the data, derived datasets, model outputs, and source code supporting the ecological niche modelling and conservation prioritization of threatened tree species in the Brazilian Atlantic Forest. The study was conducted in Espírito Santo, southeastern Brazil, using occurrence records collected between 1985 and 2023. Thirty-two threatened tree species were initially assessed for modelling eligibility, and ten species were retained after evaluating environmental niche breadth, environmental centrality, and environmental complexity. The repository includes: (1) occurrence records and processed occurrence datasets; (2) environmental predictor data and metadata; (3) correlation, clustering, and variance inflation factor results used to derive six alternative non-collinear predictor sets; (4) species-eligibility datasets; (5) MaxEnt model performance metrics, suitability predictions, and predictive-uncertainty outputs; (6) response-curve data and environmental optimum ranges used to calculate ecological stress; (7) ecological-stress and ecological-hotspot outputs; and (8) Python scripts and Google Colab notebooks used for data processing, modelling, analysis, and figure generation. The dataset supports reproducibility of the analytical workflow from occurrence-data preparation through species eligibility, environmental predictor selection, ecological niche modelling, ecological stress estimation, uncertainty assessment, and ecological hotspot identification. The data may be reused for methodological comparisons, species distribution modelling, environmental niche characterization, conservation prioritization, and development of similar workflows for threatened species in other regions. Occurrence records derived from SpeciesLink are redistributed under the CC BY 4.0 licence. External environmental datasets are identified and documented in the repository, and users should consult the original providers' terms of use and cite the corresponding data sources. No human or animal experimental data are included.