Stable oxygen isotopes (δ18O) in wood provide integrative records of plant water use and regional hydroclimatic processes, offering a powerful framework for spatial ecological analysis in tropical forests. Here, we present the first regional-scale δ18O isoscapes for Amazonian wood based on 387 trees sampled across 25 sites. After α-cellulose extraction, δ18O values were modeled using multiple linear regression (MLR) and Random Forest (RF) approaches. A Moran’s I test revealed no significant spatial autocorrelation (p = 0.73), indicating that geostatistical interpolation methods such as kriging were not appropriate for this dataset. The MLR model based on site-average data achieved an R2 of 0.70, with a mean absolute error (MAE) of 0.56‰ and root mean square error (RMSE) of 0.68‰. The RF model showed comparable performance (R2 = 0.67; MAE = 0.64‰; RMSE = 0.77‰). Both approaches reproduced a coherent southeast-to-northwest gradient, with lower δ18O values in the western Amazon and higher values in the east, consistent with regional patterns in precipitation isotopic composition and evapotranspiration. These findings demonstrate that climate-driven statistical modeling effectively captures large-scale isotopic structure across the Amazon basin, providing a robust spatial representation of δ18O variability in tropical forest wood.
Batista et al. (2026) studied this question.