Language ranges follow environmental variation in the Uralic speaker area

## Blind Review Notice This repository is anonymised for double-anonymous peer review.Citation details and authorship metadata will be added upon publication. This release provides a fully reproducible workflow for analysing ecological structure and language distributions across northern Eurasia. It integrates two complementary components: Ecozone Classification Environmental structure is defined using 15 environmental variables aggregated to a 50 km grid. The workflow includes: Principal Component Analysis (PCA) Parallel analysis and component retention diagnostics Silhouette-based clustering validation k-means ecozone classification in PCA space Ecozone mapping and summary statistics The outputs include PCA loadings, PCA scores, cluster assignments, and ecozone-level summary tables. Ecological Niche Modelling Language distributions are represented as polygon-based presence–absence data aggregated to the same grid resolution. The modelling workflow includes: Random Forest classification models per language Model evaluation (AUC, ROC, confusion matrices) Threshold optimisation Variable importance estimation Partial dependence analysis Generation of environmental suitability and probability maps All model outputs, evaluation statistics, and figure source data are provided in machine-readable format.