Variation in wood density across South American tropical forests

Predicted spatial variation in wood density across tropical South American forests. This data package contains map data products from the paper "Variation in wood density across South American tropical forests".
Contents:
Ensemble predictions.gri - Predicted wood density averaged across models. Maps shown in Fig. 3 and Fig. S2b.
RF spat.gri - Predicted wood density from random forest model - spatial covariates only. Maps shown in Fig S2a.
RF env.gri - Predicted wood density from random forest model - environmental covariates only. Map shown in Fig S2a.
RF all.gri - Predicted wood density from random forest model - environmental and spatial covariates. Map shown in Fig S2a.
GAM spat.gri - Predicted wood density from GAM - spatial covariates only. Map shown in Fig S2a.
GAM env.gri - Predicted wood density from GAM - environmental covariates only. Map shown in Fig S2a.
GAM all.gri - Predicted wood density from GAM - environmental and spatial covariates. Map shown in Fig S2a.
SD predictions.gri - Standard deviation of model predictions. Map shown in Fig. S3.
Dissimilarity index.gri - Multivariate dissimilarity index. Fig. S7b.
EnvRobustSubsampling.gri - Binary variable - are environmental variables in the range where relationships were robust to subsampling data. Fig. S7c.
EnvRangeTraining.gri - Binary variable - are environmental variables in the range seen in the training data. Fig. S7d.
AOA1.gri - Binary variable - area of model applicability defined using spatial cross-validation. Fig. S7e.
AOA2.gri - Binary variable - area of model applicability defined using non-spatial cross-validation. Fig. S7d.