 Hydrological models are widely used to characterize, understand and manage hydra systems. Lumped parameter models are of particular interest in karst environments due to their complexity and heterogeneity. Two lumped parameter modeling approaches were compared artificial neural networks, ANNs and reservoir models. Five karst systems in the Mediterranean and Alpine regions were studied using various performance criteria over different hydrological periods. Both ANNs and reservoir models were found to be effective at simulating spring discharge, although each has its own strengths and weaknesses. ANNs are more flexible than reservoir models when it comes to input data formats and amounts, while reservoir models are better suited for reproducing low flow conditions. Both models struggle to reproduce extreme events such as droughts and floods, however, which is a known issue in hydrological modeling. ANNs have been shown to be useful for identifying recharge areas and delineating catchments, while reservoir models are adapted to understanding the hydrological functioning of a system through model structure and parameters. This article was authored by G. Sinkus, Awunsch, N. Mozilli, and others. We are article.tv, links in the description below.