Research compendium for 'Machine-learning models of δ¹³C and δ¹⁵N isoscapes in Amazonian wood'

This project compiles the data, code, and final products associated with the study “Machine-learning models of δ¹³C and δ¹⁵N isoscapes in Amazonian wood”. The research investigates the spatial variability of carbon and nitrogen stable isotope ratios in Amazonian wood and evaluates their potential use as complementary tools for ecological interpretation and timber provenance assessment at the basin scale. Using isotopic measurements from 571 trees across 47 sites and a suite of environmental, climatic, and edaphic covariates, Random Forest and Quantile Regression Forest models were developed to generate predictive isoscapes and associated uncertainty maps for δ¹³C and δ¹⁵N. The repository includes the processed isotopic dataset, geospatial covariates, R scripts used for data analysis and isoscape modeling, and final map outputs. These materials are provided to support transparency, reproducibility, and future research on stable isotope applications in tropical forest ecology and timber traceability.