Species distribution modelling with Bayesian additive regression tree (BART) methods

This course develops a species-distribution modelling workflow using Bayesian additive regression trees. Participants compare BART with other modelling approaches and use the R packages embarcadero and dbarts to choose environmental predictors, train and evaluate occurrence models, and predict presence or absence in new places or periods. The programme introduces Bayesian regression and boosted trees, then examines predictor effects, interactions, spatial effects and random-intercept BART models, with practical troubleshooting and interpretation.

Live online teaching takes place from 15:00 to 18:00 Berlin time (CET, UTC+1) on 9–11 February 2027. The programme also specifies four additional hours of independent practical work using annotated R scripts and online support. Advanced students, researchers and practitioners should already know the basics of species-distribution modelling and be comfortable using RStudio, installing packages and editing simple scripts. A laptop, webcam and reliable internet connection are required. The standalone course fee is EUR 390.