Performance measures for every model built to predict bird primary lifestyle through morphological traits (n = 11 009 species).

The first three models were built using extreme gradient boosting algorithm (called xgboost models to avoid confusions), and the last two correspond to traditional approaches using the same features (all morphological traits) than the best xgboost model trained through extreme gradient bosting: multinomial logistic regression (MLR) and linear discriminant analysis (LDA). The last two models offer a contrast between traditional approaches used in classification tasks and the machine learning algorithm used in this research in front of dimensionality of features.