Carsten F. Dormann

This study evaluates process-guided neural networks (PGNNs) and standard ANNs for predicting forest carbon fluxes, showing PGNNs outperform a process model in data-sparse and transfer scenarios.

The paper shows climate, not latitude, drives plant-pollinator specialisation globally, with specialisation peaking at northern mid-latitudes and declining in warmer, more seasonal climates.

This paper combines crowdsourced biodiversity data with Earth observation to model plant traits at high resolution, achieving strong predictive power and improving spatial coverage for global trait mapping.

The paper re-analyzes Liang et al. (2016) data, finding no local-scale effect of tree species richness on productivity and highlighting regional variability.